consul/agent/consul/prepared_query_endpoint_tes...

3168 lines
92 KiB
Go
Raw Normal View History

package consul
import (
"bytes"
"fmt"
"net/rpc"
"os"
"reflect"
2015-11-11 02:23:37 +00:00
"sort"
"strings"
"testing"
"time"
"github.com/hashicorp/consul/acl"
"github.com/hashicorp/consul/agent/structs"
tokenStore "github.com/hashicorp/consul/agent/token"
"github.com/hashicorp/consul/api"
"github.com/hashicorp/consul/sdk/testutil/retry"
"github.com/hashicorp/consul/testrpc"
"github.com/hashicorp/consul/types"
"github.com/hashicorp/go-hclog"
msgpackrpc "github.com/hashicorp/net-rpc-msgpackrpc"
"github.com/hashicorp/serf/coordinate"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
)
func TestPreparedQuery_Apply(t *testing.T) {
t.Parallel()
dir1, s1 := testServer(t)
defer os.RemoveAll(dir1)
defer s1.Shutdown()
codec := rpcClient(t, s1)
defer codec.Close()
testrpc.WaitForLeader(t, s1.RPC, "dc1")
// Set up a bare bones query.
2015-11-10 19:16:17 +00:00
query := structs.PreparedQueryRequest{
Datacenter: "dc1",
Op: structs.PreparedQueryCreate,
Query: &structs.PreparedQuery{
Name: "test",
Service: structs.ServiceQuery{
Service: "redis",
},
},
}
var reply string
// Set an ID which should fail the create.
2015-11-10 19:16:17 +00:00
query.Query.ID = "nope"
err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply)
if err == nil || !strings.Contains(err.Error(), "ID must be empty") {
t.Fatalf("bad: %v", err)
}
// Change it to a bogus modify which should also fail.
2015-11-10 19:16:17 +00:00
query.Op = structs.PreparedQueryUpdate
query.Query.ID = generateUUID()
err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply)
if err == nil || !strings.Contains(err.Error(), "Cannot modify non-existent prepared query") {
t.Fatalf("bad: %v", err)
}
// Fix up the ID but invalidate the query itself. This proves we call
// parseQuery for a create, but that function is checked in detail as
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
// part of another test so we don't have to exercise all the checks
// here.
2015-11-10 19:16:17 +00:00
query.Op = structs.PreparedQueryCreate
query.Query.ID = ""
query.Query.Service.Failover.NearestN = -1
err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply)
if err == nil || !strings.Contains(err.Error(), "Bad NearestN") {
t.Fatalf("bad: %v", err)
}
// Fix that and make sure it propagates an error from the Raft apply.
2015-11-10 19:16:17 +00:00
query.Query.Service.Failover.NearestN = 0
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
query.Query.Session = "nope"
2015-11-10 19:16:17 +00:00
err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply)
if err == nil || !strings.Contains(err.Error(), "invalid session") {
t.Fatalf("bad: %v", err)
}
// Fix that and make sure the apply goes through.
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
query.Query.Session = ""
2015-11-10 19:16:17 +00:00
if err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply); err != nil {
t.Fatalf("err: %v", err)
}
// Capture the ID and read the query back to verify.
2015-11-10 19:16:17 +00:00
query.Query.ID = reply
{
req := &structs.PreparedQuerySpecificRequest{
Datacenter: "dc1",
QueryID: query.Query.ID,
}
var resp structs.IndexedPreparedQueries
if err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Get", req, &resp); err != nil {
t.Fatalf("err: %v", err)
}
if len(resp.Queries) != 1 {
t.Fatalf("bad: %v", resp)
}
actual := resp.Queries[0]
if resp.Index != actual.ModifyIndex {
t.Fatalf("bad index: %d", resp.Index)
}
actual.CreateIndex, actual.ModifyIndex = 0, 0
2015-11-10 19:16:17 +00:00
if !reflect.DeepEqual(actual, query.Query) {
t.Fatalf("bad: %v", actual)
}
}
// Make the op an update. This should go through now that we have an ID.
2015-11-10 19:16:17 +00:00
query.Op = structs.PreparedQueryUpdate
query.Query.Service.Failover.NearestN = 2
if err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply); err != nil {
t.Fatalf("err: %v", err)
}
// Read back again to verify the update worked.
{
req := &structs.PreparedQuerySpecificRequest{
Datacenter: "dc1",
QueryID: query.Query.ID,
}
var resp structs.IndexedPreparedQueries
if err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Get", req, &resp); err != nil {
t.Fatalf("err: %v", err)
}
if len(resp.Queries) != 1 {
t.Fatalf("bad: %v", resp)
}
actual := resp.Queries[0]
if resp.Index != actual.ModifyIndex {
t.Fatalf("bad index: %d", resp.Index)
}
actual.CreateIndex, actual.ModifyIndex = 0, 0
2015-11-10 19:16:17 +00:00
if !reflect.DeepEqual(actual, query.Query) {
t.Fatalf("bad: %v", actual)
}
}
// Give a bogus op and make sure it fails.
2015-11-10 19:16:17 +00:00
query.Op = "nope"
err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply)
if err == nil || !strings.Contains(err.Error(), "Unknown prepared query operation:") {
t.Fatalf("bad: %v", err)
}
// Prove that an update also goes through the parseQuery validation.
2015-11-10 19:16:17 +00:00
query.Op = structs.PreparedQueryUpdate
query.Query.Service.Failover.NearestN = -1
err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply)
if err == nil || !strings.Contains(err.Error(), "Bad NearestN") {
t.Fatalf("bad: %v", err)
}
// Now change the op to delete; the bad query field should be ignored
// because all we care about for a delete op is the ID.
2015-11-10 19:16:17 +00:00
query.Op = structs.PreparedQueryDelete
if err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply); err != nil {
t.Fatalf("err: %v", err)
}
// Verify that this query is deleted.
{
req := &structs.PreparedQuerySpecificRequest{
Datacenter: "dc1",
QueryID: query.Query.ID,
2015-11-10 19:16:17 +00:00
}
var resp structs.IndexedPreparedQueries
if err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Get", req, &resp); err != nil {
if !structs.IsErrQueryNotFound(err) {
t.Fatalf("err: %v", err)
}
2015-11-10 19:16:17 +00:00
}
if len(resp.Queries) != 0 {
t.Fatalf("bad: %v", resp)
}
}
}
func TestPreparedQuery_Apply_ACLDeny(t *testing.T) {
t.Parallel()
2015-11-10 19:16:17 +00:00
dir1, s1 := testServerWithConfig(t, func(c *Config) {
c.ACLDatacenter = "dc1"
New ACLs (#4791) This PR is almost a complete rewrite of the ACL system within Consul. It brings the features more in line with other HashiCorp products. Obviously there is quite a bit left to do here but most of it is related docs, testing and finishing the last few commands in the CLI. I will update the PR description and check off the todos as I finish them over the next few days/week. Description At a high level this PR is mainly to split ACL tokens from Policies and to split the concepts of Authorization from Identities. A lot of this PR is mostly just to support CRUD operations on ACLTokens and ACLPolicies. These in and of themselves are not particularly interesting. The bigger conceptual changes are in how tokens get resolved, how backwards compatibility is handled and the separation of policy from identity which could lead the way to allowing for alternative identity providers. On the surface and with a new cluster the ACL system will look very similar to that of Nomads. Both have tokens and policies. Both have local tokens. The ACL management APIs for both are very similar. I even ripped off Nomad's ACL bootstrap resetting procedure. There are a few key differences though. Nomad requires token and policy replication where Consul only requires policy replication with token replication being opt-in. In Consul local tokens only work with token replication being enabled though. All policies in Nomad are globally applicable. In Consul all policies are stored and replicated globally but can be scoped to a subset of the datacenters. This allows for more granular access management. Unlike Nomad, Consul has legacy baggage in the form of the original ACL system. The ramifications of this are: A server running the new system must still support other clients using the legacy system. A client running the new system must be able to use the legacy RPCs when the servers in its datacenter are running the legacy system. The primary ACL DC's servers running in legacy mode needs to be a gate that keeps everything else in the entire multi-DC cluster running in legacy mode. So not only does this PR implement the new ACL system but has a legacy mode built in for when the cluster isn't ready for new ACLs. Also detecting that new ACLs can be used is automatic and requires no configuration on the part of administrators. This process is detailed more in the "Transitioning from Legacy to New ACL Mode" section below.
2018-10-19 16:04:07 +00:00
c.ACLsEnabled = true
2015-11-10 19:16:17 +00:00
c.ACLMasterToken = "root"
c.ACLDefaultPolicy = "deny"
})
defer os.RemoveAll(dir1)
defer s1.Shutdown()
codec := rpcClient(t, s1)
defer codec.Close()
testrpc.WaitForLeader(t, s1.RPC, "dc1", testrpc.WithToken("root"))
2015-11-10 19:16:17 +00:00
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
// Create an ACL with write permissions for redis queries.
var token string
2015-11-10 19:16:17 +00:00
{
var rules = `
query "redis" {
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
policy = "write"
2015-11-10 19:16:17 +00:00
}
`
req := structs.ACLRequest{
Datacenter: "dc1",
Op: structs.ACLSet,
ACL: structs.ACL{
Name: "User token",
New ACLs (#4791) This PR is almost a complete rewrite of the ACL system within Consul. It brings the features more in line with other HashiCorp products. Obviously there is quite a bit left to do here but most of it is related docs, testing and finishing the last few commands in the CLI. I will update the PR description and check off the todos as I finish them over the next few days/week. Description At a high level this PR is mainly to split ACL tokens from Policies and to split the concepts of Authorization from Identities. A lot of this PR is mostly just to support CRUD operations on ACLTokens and ACLPolicies. These in and of themselves are not particularly interesting. The bigger conceptual changes are in how tokens get resolved, how backwards compatibility is handled and the separation of policy from identity which could lead the way to allowing for alternative identity providers. On the surface and with a new cluster the ACL system will look very similar to that of Nomads. Both have tokens and policies. Both have local tokens. The ACL management APIs for both are very similar. I even ripped off Nomad's ACL bootstrap resetting procedure. There are a few key differences though. Nomad requires token and policy replication where Consul only requires policy replication with token replication being opt-in. In Consul local tokens only work with token replication being enabled though. All policies in Nomad are globally applicable. In Consul all policies are stored and replicated globally but can be scoped to a subset of the datacenters. This allows for more granular access management. Unlike Nomad, Consul has legacy baggage in the form of the original ACL system. The ramifications of this are: A server running the new system must still support other clients using the legacy system. A client running the new system must be able to use the legacy RPCs when the servers in its datacenter are running the legacy system. The primary ACL DC's servers running in legacy mode needs to be a gate that keeps everything else in the entire multi-DC cluster running in legacy mode. So not only does this PR implement the new ACL system but has a legacy mode built in for when the cluster isn't ready for new ACLs. Also detecting that new ACLs can be used is automatic and requires no configuration on the part of administrators. This process is detailed more in the "Transitioning from Legacy to New ACL Mode" section below.
2018-10-19 16:04:07 +00:00
Type: structs.ACLTokenTypeClient,
2015-11-10 19:16:17 +00:00
Rules: rules,
},
WriteRequest: structs.WriteRequest{Token: "root"},
}
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
if err := msgpackrpc.CallWithCodec(codec, "ACL.Apply", &req, &token); err != nil {
2015-11-10 19:16:17 +00:00
t.Fatalf("err: %v", err)
}
}
// Set up a bare bones query.
query := structs.PreparedQueryRequest{
Datacenter: "dc1",
Op: structs.PreparedQueryCreate,
Query: &structs.PreparedQuery{
Name: "redis-master",
2015-11-10 19:16:17 +00:00
Service: structs.ServiceQuery{
Service: "the-redis",
2015-11-10 19:16:17 +00:00
},
},
}
var reply string
// Creating without a token should fail since the default policy is to
// deny.
err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply)
if !acl.IsErrPermissionDenied(err) {
2015-11-10 19:16:17 +00:00
t.Fatalf("bad: %v", err)
}
// Now add the token and try again.
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
query.WriteRequest.Token = token
2015-11-10 19:16:17 +00:00
if err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply); err != nil {
t.Fatalf("err: %v", err)
}
// Capture the ID and set the token, then read back the query to verify.
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
// Note that unlike previous versions of Consul, we DO NOT capture the
// token. We will set that here just to be explicit about it.
2015-11-10 19:16:17 +00:00
query.Query.ID = reply
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
query.Query.Token = ""
2015-11-10 19:16:17 +00:00
{
req := &structs.PreparedQuerySpecificRequest{
Datacenter: "dc1",
QueryID: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: "root"},
2015-11-10 19:16:17 +00:00
}
var resp structs.IndexedPreparedQueries
if err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Get", req, &resp); err != nil {
2015-11-10 19:16:17 +00:00
t.Fatalf("err: %v", err)
}
if len(resp.Queries) != 1 {
t.Fatalf("bad: %v", resp)
}
actual := resp.Queries[0]
if resp.Index != actual.ModifyIndex {
t.Fatalf("bad index: %d", resp.Index)
}
actual.CreateIndex, actual.ModifyIndex = 0, 0
if !reflect.DeepEqual(actual, query.Query) {
t.Fatalf("bad: %v", actual)
}
}
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
// Try to do an update without a token; this should get rejected.
2015-11-10 19:16:17 +00:00
query.Op = structs.PreparedQueryUpdate
query.WriteRequest.Token = ""
err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply)
if !acl.IsErrPermissionDenied(err) {
2015-11-10 19:16:17 +00:00
t.Fatalf("bad: %v", err)
}
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
// Try again with the original token; this should go through.
query.WriteRequest.Token = token
2015-11-10 19:16:17 +00:00
if err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply); err != nil {
t.Fatalf("err: %v", err)
}
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
// Try to do a delete with no token; this should get rejected.
2015-11-10 19:16:17 +00:00
query.Op = structs.PreparedQueryDelete
query.WriteRequest.Token = ""
err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply)
if !acl.IsErrPermissionDenied(err) {
2015-11-10 19:16:17 +00:00
t.Fatalf("bad: %v", err)
}
// Try again with the original token. This should go through.
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
query.WriteRequest.Token = token
2015-11-10 19:16:17 +00:00
if err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply); err != nil {
t.Fatalf("err: %v", err)
}
// Make sure the query got deleted.
{
req := &structs.PreparedQuerySpecificRequest{
Datacenter: "dc1",
QueryID: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: "root"},
2015-11-10 19:16:17 +00:00
}
var resp structs.IndexedPreparedQueries
if err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Get", req, &resp); err != nil {
if !structs.IsErrQueryNotFound(err) {
t.Fatalf("err: %v", err)
}
2015-11-10 19:16:17 +00:00
}
if len(resp.Queries) != 0 {
t.Fatalf("bad: %v", resp)
}
}
// Make the query again.
query.Op = structs.PreparedQueryCreate
query.Query.ID = ""
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
query.WriteRequest.Token = token
2015-11-10 19:16:17 +00:00
if err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply); err != nil {
t.Fatalf("err: %v", err)
}
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
// Check that it's there, and again make sure that the token did not get
// captured.
2015-11-10 19:16:17 +00:00
query.Query.ID = reply
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
query.Query.Token = ""
2015-11-10 19:16:17 +00:00
{
req := &structs.PreparedQuerySpecificRequest{
Datacenter: "dc1",
QueryID: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: "root"},
2015-11-10 19:16:17 +00:00
}
var resp structs.IndexedPreparedQueries
if err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Get", req, &resp); err != nil {
2015-11-10 19:16:17 +00:00
t.Fatalf("err: %v", err)
}
if len(resp.Queries) != 1 {
t.Fatalf("bad: %v", resp)
}
actual := resp.Queries[0]
if resp.Index != actual.ModifyIndex {
t.Fatalf("bad index: %d", resp.Index)
}
actual.CreateIndex, actual.ModifyIndex = 0, 0
if !reflect.DeepEqual(actual, query.Query) {
t.Fatalf("bad: %v", actual)
}
}
// A management token should be able to update the query no matter what.
query.Op = structs.PreparedQueryUpdate
query.WriteRequest.Token = "root"
if err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply); err != nil {
t.Fatalf("err: %v", err)
}
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
// That last update should not have captured a token.
query.Query.Token = ""
2015-11-10 19:16:17 +00:00
{
req := &structs.PreparedQuerySpecificRequest{
Datacenter: "dc1",
QueryID: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: "root"},
2015-11-10 19:16:17 +00:00
}
var resp structs.IndexedPreparedQueries
if err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Get", req, &resp); err != nil {
2015-11-10 19:16:17 +00:00
t.Fatalf("err: %v", err)
}
if len(resp.Queries) != 1 {
t.Fatalf("bad: %v", resp)
}
actual := resp.Queries[0]
if resp.Index != actual.ModifyIndex {
t.Fatalf("bad index: %d", resp.Index)
}
actual.CreateIndex, actual.ModifyIndex = 0, 0
if !reflect.DeepEqual(actual, query.Query) {
t.Fatalf("bad: %v", actual)
}
}
// A management token should be able to delete the query no matter what.
query.Op = structs.PreparedQueryDelete
query.WriteRequest.Token = "root"
if err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply); err != nil {
t.Fatalf("err: %v", err)
}
// Make sure the query got deleted.
{
req := &structs.PreparedQuerySpecificRequest{
Datacenter: "dc1",
QueryID: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: "root"},
}
var resp structs.IndexedPreparedQueries
if err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Get", req, &resp); err != nil {
if !structs.IsErrQueryNotFound(err) {
t.Fatalf("err: %v", err)
}
}
if len(resp.Queries) != 0 {
t.Fatalf("bad: %v", resp)
}
}
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
// Use the root token to make a query under a different name.
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
query.Op = structs.PreparedQueryCreate
query.Query.ID = ""
query.Query.Name = "cassandra"
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
query.WriteRequest.Token = "root"
if err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply); err != nil {
t.Fatalf("err: %v", err)
}
// Check that it's there and that the token did not get captured.
query.Query.ID = reply
query.Query.Token = ""
{
req := &structs.PreparedQuerySpecificRequest{
Datacenter: "dc1",
QueryID: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: "root"},
}
var resp structs.IndexedPreparedQueries
if err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Get", req, &resp); err != nil {
t.Fatalf("err: %v", err)
}
if len(resp.Queries) != 1 {
t.Fatalf("bad: %v", resp)
}
actual := resp.Queries[0]
if resp.Index != actual.ModifyIndex {
t.Fatalf("bad index: %d", resp.Index)
}
actual.CreateIndex, actual.ModifyIndex = 0, 0
if !reflect.DeepEqual(actual, query.Query) {
t.Fatalf("bad: %v", actual)
}
}
// Now try to change that to redis with the valid redis token. This will
// fail because that token can't change cassandra queries.
query.Op = structs.PreparedQueryUpdate
query.Query.Name = "redis"
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
query.WriteRequest.Token = token
err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply)
if !acl.IsErrPermissionDenied(err) {
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
t.Fatalf("bad: %v", err)
}
}
2015-11-10 19:33:00 +00:00
func TestPreparedQuery_Apply_ForwardLeader(t *testing.T) {
t.Parallel()
Adds support for snapshots and restores. (#2396) * Updates Raft library to get new snapshot/restore API. * Basic backup and restore working, but need some cleanup. * Breaks out a snapshot module and adds a SHA256 integrity check. * Adds snapshot ACL and fills in some missing comments. * Require a consistent read for snapshots. * Make sure snapshot works if ACLs aren't enabled. * Adds a bit of package documentation. * Returns an empty response from restore to avoid EOF errors. * Adds API client support for snapshots. * Makes internal file names match on-disk file snapshots. * Adds DC and token coverage for snapshot API test. * Adds missing documentation. * Adds a unit test for the snapshot client endpoint. * Moves the connection pool out of the client for easier testing. * Fixes an incidental issue in the prepared query unit test. I realized I had two servers in bootstrap mode so this wasn't a good setup. * Adds a half close to the TCP stream and fixes panic on error. * Adds client and endpoint tests for snapshots. * Moves the pool back into the snapshot RPC client. * Adds a TLS test and fixes half-closes for TLS connections. * Tweaks some comments. * Adds a low-level snapshot test. This is independent of Consul so we can pull this out into a library later if we want to. * Cleans up snapshot and archive and completes archive tests. * Sends a clear error for snapshot operations in dev mode. Snapshots require the Raft snapshots to be readable, which isn't supported in dev mode. Send a clear error instead of a deep-down Raft one. * Adds docs for the snapshot endpoint. * Adds a stale mode and index feedback for snapshot saves. This gives folks a way to extract data even if the cluster has no leader. * Changes the internal format of a snapshot from zip to tgz. * Pulls in Raft fix to cancel inflight before a restore. * Pulls in new Raft restore interface. * Adds metadata to snapshot saves and a verify function. * Adds basic save and restore snapshot CLI commands. * Gets rid of tarball extensions and adds restore message. * Fixes an incidental bad link in the KV docs. * Adds documentation for the snapshot CLI commands. * Scuttle any request body when a snapshot is saved. * Fixes archive unit test error message check. * Allows for nil output writers in snapshot RPC handlers. * Renames hash list Decode to DecodeAndVerify. * Closes the client connection for snapshot ops. * Lowers timeout for restore ops. * Updates Raft vendor to get new Restore signature and integrates with Consul. * Bounces the leader's internal state when we do a restore.
2016-10-26 02:20:24 +00:00
dir1, s1 := testServerWithConfig(t, func(c *Config) {
c.Bootstrap = false
})
defer os.RemoveAll(dir1)
defer s1.Shutdown()
codec1 := rpcClient(t, s1)
defer codec1.Close()
dir2, s2 := testServer(t)
defer os.RemoveAll(dir2)
defer s2.Shutdown()
codec2 := rpcClient(t, s2)
defer codec2.Close()
// Try to join.
2017-05-05 10:29:49 +00:00
joinLAN(t, s2, s1)
testrpc.WaitForLeader(t, s1.RPC, "dc1")
testrpc.WaitForLeader(t, s2.RPC, "dc1")
// Use the follower as the client.
var codec rpc.ClientCodec
if !s1.IsLeader() {
codec = codec1
} else {
codec = codec2
}
// Set up a node and service in the catalog.
{
req := structs.RegisterRequest{
Datacenter: "dc1",
Node: "foo",
Address: "127.0.0.1",
Service: &structs.NodeService{
Service: "redis",
Tags: []string{"master"},
Port: 8000,
},
}
var reply struct{}
err := msgpackrpc.CallWithCodec(codec, "Catalog.Register", &req, &reply)
if err != nil {
t.Fatalf("err: %v", err)
}
}
// Set up a bare bones query.
query := structs.PreparedQueryRequest{
Datacenter: "dc1",
Op: structs.PreparedQueryCreate,
Query: &structs.PreparedQuery{
Name: "test",
Service: structs.ServiceQuery{
Service: "redis",
},
},
}
// Make sure the apply works even when forwarded through the non-leader.
var reply string
if err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply); err != nil {
t.Fatalf("err: %v", err)
}
}
2015-11-10 19:33:00 +00:00
func TestPreparedQuery_parseQuery(t *testing.T) {
t.Parallel()
2015-11-10 19:33:00 +00:00
query := &structs.PreparedQuery{}
err := parseQuery(query)
if err == nil || !strings.Contains(err.Error(), "Must be bound to a session") {
2015-11-10 19:33:00 +00:00
t.Fatalf("bad: %v", err)
}
query.Session = "adf4238a-882b-9ddc-4a9d-5b6758e4159e"
err = parseQuery(query)
if err == nil || !strings.Contains(err.Error(), "Must provide a Service") {
t.Fatalf("bad: %v", err)
}
query.Session = ""
query.Template.Type = "some-kind-of-template"
err = parseQuery(query)
if err == nil || !strings.Contains(err.Error(), "Must provide a Service") {
t.Fatalf("bad: %v", err)
}
query.Template.Type = ""
err = parseQuery(query)
if err == nil || !strings.Contains(err.Error(), "Must be bound to a session") {
2015-11-10 19:33:00 +00:00
t.Fatalf("bad: %v", err)
}
// None of the rest of these care about version 8 ACL enforcement.
query = &structs.PreparedQuery{}
query.Session = "adf4238a-882b-9ddc-4a9d-5b6758e4159e"
query.Service.Service = "foo"
if err := parseQuery(query); err != nil {
t.Fatalf("err: %v", err)
}
2015-11-10 19:33:00 +00:00
query.Token = redactedToken
err = parseQuery(query)
if err == nil || !strings.Contains(err.Error(), "Bad Token") {
t.Fatalf("bad: %v", err)
}
2015-11-10 19:33:00 +00:00
query.Token = "adf4238a-882b-9ddc-4a9d-5b6758e4159e"
if err := parseQuery(query); err != nil {
t.Fatalf("err: %v", err)
}
2015-11-10 19:33:00 +00:00
query.Service.Failover.NearestN = -1
err = parseQuery(query)
if err == nil || !strings.Contains(err.Error(), "Bad NearestN") {
t.Fatalf("bad: %v", err)
}
query.Service.Failover.NearestN = 3
if err := parseQuery(query); err != nil {
t.Fatalf("err: %v", err)
}
query.DNS.TTL = "two fortnights"
err = parseQuery(query)
if err == nil || !strings.Contains(err.Error(), "Bad DNS TTL") {
t.Fatalf("bad: %v", err)
}
query.DNS.TTL = "-3s"
err = parseQuery(query)
if err == nil || !strings.Contains(err.Error(), "must be >=0") {
t.Fatalf("bad: %v", err)
}
query.DNS.TTL = "3s"
if err := parseQuery(query); err != nil {
t.Fatalf("err: %v", err)
}
query.Service.NodeMeta = map[string]string{"": "somevalue"}
err = parseQuery(query)
if err == nil || !strings.Contains(err.Error(), "cannot be blank") {
t.Fatalf("bad: %v", err)
}
query.Service.NodeMeta = map[string]string{"somekey": "somevalue"}
if err := parseQuery(query); err != nil {
t.Fatalf("err: %v", err)
2015-11-10 19:33:00 +00:00
}
}
2016-03-03 09:04:12 +00:00
func TestPreparedQuery_ACLDeny_Catchall_Template(t *testing.T) {
t.Parallel()
dir1, s1 := testServerWithConfig(t, func(c *Config) {
c.ACLDatacenter = "dc1"
New ACLs (#4791) This PR is almost a complete rewrite of the ACL system within Consul. It brings the features more in line with other HashiCorp products. Obviously there is quite a bit left to do here but most of it is related docs, testing and finishing the last few commands in the CLI. I will update the PR description and check off the todos as I finish them over the next few days/week. Description At a high level this PR is mainly to split ACL tokens from Policies and to split the concepts of Authorization from Identities. A lot of this PR is mostly just to support CRUD operations on ACLTokens and ACLPolicies. These in and of themselves are not particularly interesting. The bigger conceptual changes are in how tokens get resolved, how backwards compatibility is handled and the separation of policy from identity which could lead the way to allowing for alternative identity providers. On the surface and with a new cluster the ACL system will look very similar to that of Nomads. Both have tokens and policies. Both have local tokens. The ACL management APIs for both are very similar. I even ripped off Nomad's ACL bootstrap resetting procedure. There are a few key differences though. Nomad requires token and policy replication where Consul only requires policy replication with token replication being opt-in. In Consul local tokens only work with token replication being enabled though. All policies in Nomad are globally applicable. In Consul all policies are stored and replicated globally but can be scoped to a subset of the datacenters. This allows for more granular access management. Unlike Nomad, Consul has legacy baggage in the form of the original ACL system. The ramifications of this are: A server running the new system must still support other clients using the legacy system. A client running the new system must be able to use the legacy RPCs when the servers in its datacenter are running the legacy system. The primary ACL DC's servers running in legacy mode needs to be a gate that keeps everything else in the entire multi-DC cluster running in legacy mode. So not only does this PR implement the new ACL system but has a legacy mode built in for when the cluster isn't ready for new ACLs. Also detecting that new ACLs can be used is automatic and requires no configuration on the part of administrators. This process is detailed more in the "Transitioning from Legacy to New ACL Mode" section below.
2018-10-19 16:04:07 +00:00
c.ACLsEnabled = true
c.ACLMasterToken = "root"
c.ACLDefaultPolicy = "deny"
})
defer os.RemoveAll(dir1)
defer s1.Shutdown()
codec := rpcClient(t, s1)
defer codec.Close()
testrpc.WaitForLeader(t, s1.RPC, "dc1", testrpc.WithToken("root"))
// Create an ACL with write permissions for any prefix.
var token string
{
var rules = `
query "" {
policy = "write"
}
`
req := structs.ACLRequest{
Datacenter: "dc1",
Op: structs.ACLSet,
ACL: structs.ACL{
Name: "User token",
New ACLs (#4791) This PR is almost a complete rewrite of the ACL system within Consul. It brings the features more in line with other HashiCorp products. Obviously there is quite a bit left to do here but most of it is related docs, testing and finishing the last few commands in the CLI. I will update the PR description and check off the todos as I finish them over the next few days/week. Description At a high level this PR is mainly to split ACL tokens from Policies and to split the concepts of Authorization from Identities. A lot of this PR is mostly just to support CRUD operations on ACLTokens and ACLPolicies. These in and of themselves are not particularly interesting. The bigger conceptual changes are in how tokens get resolved, how backwards compatibility is handled and the separation of policy from identity which could lead the way to allowing for alternative identity providers. On the surface and with a new cluster the ACL system will look very similar to that of Nomads. Both have tokens and policies. Both have local tokens. The ACL management APIs for both are very similar. I even ripped off Nomad's ACL bootstrap resetting procedure. There are a few key differences though. Nomad requires token and policy replication where Consul only requires policy replication with token replication being opt-in. In Consul local tokens only work with token replication being enabled though. All policies in Nomad are globally applicable. In Consul all policies are stored and replicated globally but can be scoped to a subset of the datacenters. This allows for more granular access management. Unlike Nomad, Consul has legacy baggage in the form of the original ACL system. The ramifications of this are: A server running the new system must still support other clients using the legacy system. A client running the new system must be able to use the legacy RPCs when the servers in its datacenter are running the legacy system. The primary ACL DC's servers running in legacy mode needs to be a gate that keeps everything else in the entire multi-DC cluster running in legacy mode. So not only does this PR implement the new ACL system but has a legacy mode built in for when the cluster isn't ready for new ACLs. Also detecting that new ACLs can be used is automatic and requires no configuration on the part of administrators. This process is detailed more in the "Transitioning from Legacy to New ACL Mode" section below.
2018-10-19 16:04:07 +00:00
Type: structs.ACLTokenTypeClient,
Rules: rules,
},
WriteRequest: structs.WriteRequest{Token: "root"},
}
if err := msgpackrpc.CallWithCodec(codec, "ACL.Apply", &req, &token); err != nil {
t.Fatalf("err: %v", err)
}
}
// Set up a catch-all template.
query := structs.PreparedQueryRequest{
Datacenter: "dc1",
Op: structs.PreparedQueryCreate,
Query: &structs.PreparedQuery{
Name: "",
Token: "5e1e24e5-1329-f86f-18c6-3d3734edb2cd",
Template: structs.QueryTemplateOptions{
Type: structs.QueryTemplateTypeNamePrefixMatch,
},
Service: structs.ServiceQuery{
Service: "${name.full}",
},
},
}
var reply string
// Creating without a token should fail since the default policy is to
// deny.
err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply)
if !acl.IsErrPermissionDenied(err) {
t.Fatalf("bad: %v", err)
}
// Now add the token and try again.
query.WriteRequest.Token = token
if err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply); err != nil {
t.Fatalf("err: %v", err)
}
// Capture the ID and read back the query to verify. Note that the token
// will be redacted since this isn't a management token.
query.Query.ID = reply
query.Query.Token = redactedToken
{
req := &structs.PreparedQuerySpecificRequest{
Datacenter: "dc1",
QueryID: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: token},
}
var resp structs.IndexedPreparedQueries
if err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Get", req, &resp); err != nil {
t.Fatalf("err: %v", err)
}
if len(resp.Queries) != 1 {
t.Fatalf("bad: %v", resp)
}
actual := resp.Queries[0]
if resp.Index != actual.ModifyIndex {
t.Fatalf("bad index: %d", resp.Index)
}
actual.CreateIndex, actual.ModifyIndex = 0, 0
if !reflect.DeepEqual(actual, query.Query) {
t.Fatalf("bad: %v", actual)
}
}
// Try to query by ID without a token and make sure it gets denied, even
// though this has an empty name and would normally be shown.
{
req := &structs.PreparedQuerySpecificRequest{
Datacenter: "dc1",
QueryID: query.Query.ID,
}
var resp structs.IndexedPreparedQueries
err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Get", req, &resp)
if !acl.IsErrPermissionDenied(err) {
t.Fatalf("bad: %v", err)
}
if len(resp.Queries) != 0 {
t.Fatalf("bad: %v", resp)
}
}
// We should get the same result listing all the queries without a
// token.
{
req := &structs.DCSpecificRequest{
Datacenter: "dc1",
}
var resp structs.IndexedPreparedQueries
if err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.List", req, &resp); err != nil {
t.Fatalf("err: %v", err)
}
if len(resp.Queries) != 0 {
t.Fatalf("bad: %v", resp)
}
}
// But a management token should be able to see it, and the real token.
query.Query.Token = "5e1e24e5-1329-f86f-18c6-3d3734edb2cd"
{
req := &structs.PreparedQuerySpecificRequest{
Datacenter: "dc1",
QueryID: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: "root"},
}
var resp structs.IndexedPreparedQueries
if err = msgpackrpc.CallWithCodec(codec, "PreparedQuery.Get", req, &resp); err != nil {
t.Fatalf("err: %v", err)
}
if len(resp.Queries) != 1 {
t.Fatalf("bad: %v", resp)
}
actual := resp.Queries[0]
if resp.Index != actual.ModifyIndex {
t.Fatalf("bad index: %d", resp.Index)
}
actual.CreateIndex, actual.ModifyIndex = 0, 0
if !reflect.DeepEqual(actual, query.Query) {
t.Fatalf("bad: %v", actual)
}
}
2016-03-03 09:04:12 +00:00
// Explaining should also be denied without a token.
2016-03-03 09:04:12 +00:00
{
req := &structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: "anything",
}
var resp structs.PreparedQueryExplainResponse
err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Explain", req, &resp)
if !acl.IsErrPermissionDenied(err) {
2016-03-03 09:04:12 +00:00
t.Fatalf("bad: %v", err)
}
}
// The user can explain and see the redacted token.
2016-03-03 09:04:12 +00:00
query.Query.Token = redactedToken
query.Query.Service.Service = "anything"
{
req := &structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: "anything",
QueryOptions: structs.QueryOptions{Token: token},
}
var resp structs.PreparedQueryExplainResponse
err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Explain", req, &resp)
2016-03-03 09:04:12 +00:00
if err != nil {
t.Fatalf("err: %v", err)
}
actual := &resp.Query
actual.CreateIndex, actual.ModifyIndex = 0, 0
if !reflect.DeepEqual(actual, query.Query) {
t.Fatalf("bad: %v", actual)
}
}
// Make sure the management token can also explain and see the token.
2016-03-03 09:04:12 +00:00
query.Query.Token = "5e1e24e5-1329-f86f-18c6-3d3734edb2cd"
query.Query.Service.Service = "anything"
{
req := &structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: "anything",
QueryOptions: structs.QueryOptions{Token: "root"},
}
var resp structs.PreparedQueryExplainResponse
err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Explain", req, &resp)
2016-03-03 09:04:12 +00:00
if err != nil {
t.Fatalf("err: %v", err)
}
actual := &resp.Query
actual.CreateIndex, actual.ModifyIndex = 0, 0
if !reflect.DeepEqual(actual, query.Query) {
t.Fatalf("bad: %v", actual)
}
}
}
func TestPreparedQuery_Get(t *testing.T) {
t.Parallel()
dir1, s1 := testServerWithConfig(t, func(c *Config) {
c.ACLDatacenter = "dc1"
New ACLs (#4791) This PR is almost a complete rewrite of the ACL system within Consul. It brings the features more in line with other HashiCorp products. Obviously there is quite a bit left to do here but most of it is related docs, testing and finishing the last few commands in the CLI. I will update the PR description and check off the todos as I finish them over the next few days/week. Description At a high level this PR is mainly to split ACL tokens from Policies and to split the concepts of Authorization from Identities. A lot of this PR is mostly just to support CRUD operations on ACLTokens and ACLPolicies. These in and of themselves are not particularly interesting. The bigger conceptual changes are in how tokens get resolved, how backwards compatibility is handled and the separation of policy from identity which could lead the way to allowing for alternative identity providers. On the surface and with a new cluster the ACL system will look very similar to that of Nomads. Both have tokens and policies. Both have local tokens. The ACL management APIs for both are very similar. I even ripped off Nomad's ACL bootstrap resetting procedure. There are a few key differences though. Nomad requires token and policy replication where Consul only requires policy replication with token replication being opt-in. In Consul local tokens only work with token replication being enabled though. All policies in Nomad are globally applicable. In Consul all policies are stored and replicated globally but can be scoped to a subset of the datacenters. This allows for more granular access management. Unlike Nomad, Consul has legacy baggage in the form of the original ACL system. The ramifications of this are: A server running the new system must still support other clients using the legacy system. A client running the new system must be able to use the legacy RPCs when the servers in its datacenter are running the legacy system. The primary ACL DC's servers running in legacy mode needs to be a gate that keeps everything else in the entire multi-DC cluster running in legacy mode. So not only does this PR implement the new ACL system but has a legacy mode built in for when the cluster isn't ready for new ACLs. Also detecting that new ACLs can be used is automatic and requires no configuration on the part of administrators. This process is detailed more in the "Transitioning from Legacy to New ACL Mode" section below.
2018-10-19 16:04:07 +00:00
c.ACLsEnabled = true
c.ACLMasterToken = "root"
c.ACLDefaultPolicy = "deny"
})
defer os.RemoveAll(dir1)
defer s1.Shutdown()
codec := rpcClient(t, s1)
defer codec.Close()
testrpc.WaitForTestAgent(t, s1.RPC, "dc1", testrpc.WithToken("root"))
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
// Create an ACL with write permissions for redis queries.
var token string
{
var rules = `
query "redis" {
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
policy = "write"
}
`
req := structs.ACLRequest{
Datacenter: "dc1",
Op: structs.ACLSet,
ACL: structs.ACL{
Name: "User token",
New ACLs (#4791) This PR is almost a complete rewrite of the ACL system within Consul. It brings the features more in line with other HashiCorp products. Obviously there is quite a bit left to do here but most of it is related docs, testing and finishing the last few commands in the CLI. I will update the PR description and check off the todos as I finish them over the next few days/week. Description At a high level this PR is mainly to split ACL tokens from Policies and to split the concepts of Authorization from Identities. A lot of this PR is mostly just to support CRUD operations on ACLTokens and ACLPolicies. These in and of themselves are not particularly interesting. The bigger conceptual changes are in how tokens get resolved, how backwards compatibility is handled and the separation of policy from identity which could lead the way to allowing for alternative identity providers. On the surface and with a new cluster the ACL system will look very similar to that of Nomads. Both have tokens and policies. Both have local tokens. The ACL management APIs for both are very similar. I even ripped off Nomad's ACL bootstrap resetting procedure. There are a few key differences though. Nomad requires token and policy replication where Consul only requires policy replication with token replication being opt-in. In Consul local tokens only work with token replication being enabled though. All policies in Nomad are globally applicable. In Consul all policies are stored and replicated globally but can be scoped to a subset of the datacenters. This allows for more granular access management. Unlike Nomad, Consul has legacy baggage in the form of the original ACL system. The ramifications of this are: A server running the new system must still support other clients using the legacy system. A client running the new system must be able to use the legacy RPCs when the servers in its datacenter are running the legacy system. The primary ACL DC's servers running in legacy mode needs to be a gate that keeps everything else in the entire multi-DC cluster running in legacy mode. So not only does this PR implement the new ACL system but has a legacy mode built in for when the cluster isn't ready for new ACLs. Also detecting that new ACLs can be used is automatic and requires no configuration on the part of administrators. This process is detailed more in the "Transitioning from Legacy to New ACL Mode" section below.
2018-10-19 16:04:07 +00:00
Type: structs.ACLTokenTypeClient,
Rules: rules,
},
WriteRequest: structs.WriteRequest{Token: "root"},
}
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
if err := msgpackrpc.CallWithCodec(codec, "ACL.Apply", &req, &token); err != nil {
t.Fatalf("err: %v", err)
}
}
// Set up a bare bones query.
query := structs.PreparedQueryRequest{
Datacenter: "dc1",
Op: structs.PreparedQueryCreate,
Query: &structs.PreparedQuery{
Name: "redis-master",
Service: structs.ServiceQuery{
Service: "the-redis",
},
},
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
WriteRequest: structs.WriteRequest{Token: token},
}
var reply string
if err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply); err != nil {
t.Fatalf("err: %v", err)
}
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
// Capture the ID, then read back the query to verify.
query.Query.ID = reply
{
req := &structs.PreparedQuerySpecificRequest{
Datacenter: "dc1",
QueryID: query.Query.ID,
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
QueryOptions: structs.QueryOptions{Token: token},
}
var resp structs.IndexedPreparedQueries
if err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Get", req, &resp); err != nil {
t.Fatalf("err: %v", err)
}
if len(resp.Queries) != 1 {
t.Fatalf("bad: %v", resp)
}
actual := resp.Queries[0]
if resp.Index != actual.ModifyIndex {
t.Fatalf("bad index: %d", resp.Index)
}
actual.CreateIndex, actual.ModifyIndex = 0, 0
if !reflect.DeepEqual(actual, query.Query) {
t.Fatalf("bad: %v", actual)
}
}
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
// Try again with no token, which should return an error.
{
req := &structs.PreparedQuerySpecificRequest{
Datacenter: "dc1",
QueryID: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: ""},
}
var resp structs.IndexedPreparedQueries
err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Get", req, &resp)
if !acl.IsErrPermissionDenied(err) {
t.Fatalf("bad: %v", err)
}
if len(resp.Queries) != 0 {
t.Fatalf("bad: %v", resp)
}
}
// A management token should be able to read no matter what.
{
req := &structs.PreparedQuerySpecificRequest{
Datacenter: "dc1",
QueryID: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: "root"},
}
var resp structs.IndexedPreparedQueries
if err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Get", req, &resp); err != nil {
t.Fatalf("err: %v", err)
}
if len(resp.Queries) != 1 {
t.Fatalf("bad: %v", resp)
}
actual := resp.Queries[0]
if resp.Index != actual.ModifyIndex {
t.Fatalf("bad index: %d", resp.Index)
}
actual.CreateIndex, actual.ModifyIndex = 0, 0
if !reflect.DeepEqual(actual, query.Query) {
t.Fatalf("bad: %v", actual)
}
}
// Create a session.
var session string
{
req := structs.SessionRequest{
Datacenter: "dc1",
Op: structs.SessionCreate,
Session: structs.Session{
Node: s1.config.NodeName,
},
WriteRequest: structs.WriteRequest{Token: "root"},
}
if err := msgpackrpc.CallWithCodec(codec, "Session.Apply", &req, &session); err != nil {
t.Fatalf("err: %v", err)
}
}
// Now update the query to take away its name.
query.Op = structs.PreparedQueryUpdate
query.Query.Name = ""
query.Query.Session = session
if err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply); err != nil {
t.Fatalf("err: %v", err)
}
// Try again with no token, this should work since this query is only
// managed by an ID (no name) so no ACLs apply to it.
{
req := &structs.PreparedQuerySpecificRequest{
Datacenter: "dc1",
QueryID: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: ""},
}
var resp structs.IndexedPreparedQueries
if err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Get", req, &resp); err != nil {
t.Fatalf("err: %v", err)
}
if len(resp.Queries) != 1 {
t.Fatalf("bad: %v", resp)
}
actual := resp.Queries[0]
if resp.Index != actual.ModifyIndex {
t.Fatalf("bad index: %d", resp.Index)
}
actual.CreateIndex, actual.ModifyIndex = 0, 0
if !reflect.DeepEqual(actual, query.Query) {
t.Fatalf("bad: %v", actual)
}
}
// Capture a token.
query.Op = structs.PreparedQueryUpdate
query.Query.Token = "le-token"
if err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply); err != nil {
t.Fatalf("err: %v", err)
}
// This should get redacted when we read it back without a token.
query.Query.Token = redactedToken
{
req := &structs.PreparedQuerySpecificRequest{
Datacenter: "dc1",
QueryID: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: ""},
}
var resp structs.IndexedPreparedQueries
if err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Get", req, &resp); err != nil {
t.Fatalf("err: %v", err)
}
if len(resp.Queries) != 1 {
t.Fatalf("bad: %v", resp)
}
actual := resp.Queries[0]
if resp.Index != actual.ModifyIndex {
t.Fatalf("bad index: %d", resp.Index)
}
actual.CreateIndex, actual.ModifyIndex = 0, 0
if !reflect.DeepEqual(actual, query.Query) {
t.Fatalf("bad: %v", actual)
}
}
// But a management token should be able to see it.
query.Query.Token = "le-token"
{
req := &structs.PreparedQuerySpecificRequest{
Datacenter: "dc1",
QueryID: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: "root"},
}
var resp structs.IndexedPreparedQueries
if err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Get", req, &resp); err != nil {
t.Fatalf("err: %v", err)
}
if len(resp.Queries) != 1 {
t.Fatalf("bad: %v", resp)
}
actual := resp.Queries[0]
if resp.Index != actual.ModifyIndex {
t.Fatalf("bad index: %d", resp.Index)
}
actual.CreateIndex, actual.ModifyIndex = 0, 0
if !reflect.DeepEqual(actual, query.Query) {
t.Fatalf("bad: %v", actual)
}
}
// Try to get an unknown ID.
{
req := &structs.PreparedQuerySpecificRequest{
Datacenter: "dc1",
QueryID: generateUUID(),
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
QueryOptions: structs.QueryOptions{Token: token},
}
var resp structs.IndexedPreparedQueries
if err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Get", req, &resp); err != nil {
if !structs.IsErrQueryNotFound(err) {
t.Fatalf("err: %v", err)
}
}
if len(resp.Queries) != 0 {
t.Fatalf("bad: %v", resp)
}
}
}
func TestPreparedQuery_List(t *testing.T) {
t.Parallel()
dir1, s1 := testServerWithConfig(t, func(c *Config) {
c.ACLDatacenter = "dc1"
New ACLs (#4791) This PR is almost a complete rewrite of the ACL system within Consul. It brings the features more in line with other HashiCorp products. Obviously there is quite a bit left to do here but most of it is related docs, testing and finishing the last few commands in the CLI. I will update the PR description and check off the todos as I finish them over the next few days/week. Description At a high level this PR is mainly to split ACL tokens from Policies and to split the concepts of Authorization from Identities. A lot of this PR is mostly just to support CRUD operations on ACLTokens and ACLPolicies. These in and of themselves are not particularly interesting. The bigger conceptual changes are in how tokens get resolved, how backwards compatibility is handled and the separation of policy from identity which could lead the way to allowing for alternative identity providers. On the surface and with a new cluster the ACL system will look very similar to that of Nomads. Both have tokens and policies. Both have local tokens. The ACL management APIs for both are very similar. I even ripped off Nomad's ACL bootstrap resetting procedure. There are a few key differences though. Nomad requires token and policy replication where Consul only requires policy replication with token replication being opt-in. In Consul local tokens only work with token replication being enabled though. All policies in Nomad are globally applicable. In Consul all policies are stored and replicated globally but can be scoped to a subset of the datacenters. This allows for more granular access management. Unlike Nomad, Consul has legacy baggage in the form of the original ACL system. The ramifications of this are: A server running the new system must still support other clients using the legacy system. A client running the new system must be able to use the legacy RPCs when the servers in its datacenter are running the legacy system. The primary ACL DC's servers running in legacy mode needs to be a gate that keeps everything else in the entire multi-DC cluster running in legacy mode. So not only does this PR implement the new ACL system but has a legacy mode built in for when the cluster isn't ready for new ACLs. Also detecting that new ACLs can be used is automatic and requires no configuration on the part of administrators. This process is detailed more in the "Transitioning from Legacy to New ACL Mode" section below.
2018-10-19 16:04:07 +00:00
c.ACLsEnabled = true
c.ACLMasterToken = "root"
c.ACLDefaultPolicy = "deny"
})
defer os.RemoveAll(dir1)
defer s1.Shutdown()
codec := rpcClient(t, s1)
defer codec.Close()
testrpc.WaitForTestAgent(t, s1.RPC, "dc1", testrpc.WithToken("root"))
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
// Create an ACL with write permissions for redis queries.
var token string
{
var rules = `
query "redis" {
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
policy = "write"
}
`
req := structs.ACLRequest{
Datacenter: "dc1",
Op: structs.ACLSet,
ACL: structs.ACL{
Name: "User token",
New ACLs (#4791) This PR is almost a complete rewrite of the ACL system within Consul. It brings the features more in line with other HashiCorp products. Obviously there is quite a bit left to do here but most of it is related docs, testing and finishing the last few commands in the CLI. I will update the PR description and check off the todos as I finish them over the next few days/week. Description At a high level this PR is mainly to split ACL tokens from Policies and to split the concepts of Authorization from Identities. A lot of this PR is mostly just to support CRUD operations on ACLTokens and ACLPolicies. These in and of themselves are not particularly interesting. The bigger conceptual changes are in how tokens get resolved, how backwards compatibility is handled and the separation of policy from identity which could lead the way to allowing for alternative identity providers. On the surface and with a new cluster the ACL system will look very similar to that of Nomads. Both have tokens and policies. Both have local tokens. The ACL management APIs for both are very similar. I even ripped off Nomad's ACL bootstrap resetting procedure. There are a few key differences though. Nomad requires token and policy replication where Consul only requires policy replication with token replication being opt-in. In Consul local tokens only work with token replication being enabled though. All policies in Nomad are globally applicable. In Consul all policies are stored and replicated globally but can be scoped to a subset of the datacenters. This allows for more granular access management. Unlike Nomad, Consul has legacy baggage in the form of the original ACL system. The ramifications of this are: A server running the new system must still support other clients using the legacy system. A client running the new system must be able to use the legacy RPCs when the servers in its datacenter are running the legacy system. The primary ACL DC's servers running in legacy mode needs to be a gate that keeps everything else in the entire multi-DC cluster running in legacy mode. So not only does this PR implement the new ACL system but has a legacy mode built in for when the cluster isn't ready for new ACLs. Also detecting that new ACLs can be used is automatic and requires no configuration on the part of administrators. This process is detailed more in the "Transitioning from Legacy to New ACL Mode" section below.
2018-10-19 16:04:07 +00:00
Type: structs.ACLTokenTypeClient,
Rules: rules,
},
WriteRequest: structs.WriteRequest{Token: "root"},
}
if err := msgpackrpc.CallWithCodec(codec, "ACL.Apply", &req, &token); err != nil {
t.Fatalf("err: %v", err)
}
}
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
// Query with a legit token but no queries.
{
req := &structs.DCSpecificRequest{
Datacenter: "dc1",
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
QueryOptions: structs.QueryOptions{Token: token},
}
var resp structs.IndexedPreparedQueries
if err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.List", req, &resp); err != nil {
t.Fatalf("err: %v", err)
}
if len(resp.Queries) != 0 {
t.Fatalf("bad: %v", resp)
}
}
// Set up a bare bones query.
query := structs.PreparedQueryRequest{
Datacenter: "dc1",
Op: structs.PreparedQueryCreate,
Query: &structs.PreparedQuery{
Name: "redis-master",
Token: "le-token",
Service: structs.ServiceQuery{
Service: "the-redis",
},
},
WriteRequest: structs.WriteRequest{Token: token},
}
var reply string
if err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply); err != nil {
t.Fatalf("err: %v", err)
}
// Capture the ID and read back the query to verify. We also make sure
// the captured token gets redacted.
query.Query.ID = reply
query.Query.Token = redactedToken
{
req := &structs.DCSpecificRequest{
Datacenter: "dc1",
QueryOptions: structs.QueryOptions{Token: token},
}
var resp structs.IndexedPreparedQueries
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
if err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.List", req, &resp); err != nil {
t.Fatalf("err: %v", err)
}
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
if len(resp.Queries) != 1 {
t.Fatalf("bad: %v", resp)
}
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
actual := resp.Queries[0]
if resp.Index != actual.ModifyIndex {
t.Fatalf("bad index: %d", resp.Index)
}
actual.CreateIndex, actual.ModifyIndex = 0, 0
if !reflect.DeepEqual(actual, query.Query) {
t.Fatalf("bad: %v", actual)
}
}
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
// An empty token should result in an empty list because of ACL
// filtering.
{
req := &structs.DCSpecificRequest{
Datacenter: "dc1",
QueryOptions: structs.QueryOptions{Token: ""},
}
var resp structs.IndexedPreparedQueries
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
if err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.List", req, &resp); err != nil {
t.Fatalf("err: %v", err)
}
if len(resp.Queries) != 0 {
t.Fatalf("bad: %v", resp)
}
}
// But a management token should work, and be able to see the captured
// token.
query.Query.Token = "le-token"
{
req := &structs.DCSpecificRequest{
Datacenter: "dc1",
QueryOptions: structs.QueryOptions{Token: "root"},
}
var resp structs.IndexedPreparedQueries
if err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.List", req, &resp); err != nil {
t.Fatalf("err: %v", err)
}
if len(resp.Queries) != 1 {
t.Fatalf("bad: %v", resp)
}
actual := resp.Queries[0]
if resp.Index != actual.ModifyIndex {
t.Fatalf("bad index: %d", resp.Index)
}
actual.CreateIndex, actual.ModifyIndex = 0, 0
if !reflect.DeepEqual(actual, query.Query) {
t.Fatalf("bad: %v", actual)
}
}
// Create a session.
var session string
{
req := structs.SessionRequest{
Datacenter: "dc1",
Op: structs.SessionCreate,
Session: structs.Session{
Node: s1.config.NodeName,
},
WriteRequest: structs.WriteRequest{Token: "root"},
}
if err := msgpackrpc.CallWithCodec(codec, "Session.Apply", &req, &session); err != nil {
t.Fatalf("err: %v", err)
}
}
// Now take away the query name.
query.Op = structs.PreparedQueryUpdate
query.Query.Name = ""
query.Query.Session = session
if err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply); err != nil {
t.Fatalf("err: %v", err)
}
// A query with the redis token shouldn't show anything since it doesn't
// match any un-named queries.
{
req := &structs.DCSpecificRequest{
Datacenter: "dc1",
QueryOptions: structs.QueryOptions{Token: token},
}
var resp structs.IndexedPreparedQueries
if err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.List", req, &resp); err != nil {
t.Fatalf("err: %v", err)
}
if len(resp.Queries) != 0 {
t.Fatalf("bad: %v", resp)
}
}
// But a management token should work.
{
req := &structs.DCSpecificRequest{
Datacenter: "dc1",
QueryOptions: structs.QueryOptions{Token: "root"},
}
var resp structs.IndexedPreparedQueries
if err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.List", req, &resp); err != nil {
t.Fatalf("err: %v", err)
}
if len(resp.Queries) != 1 {
t.Fatalf("bad: %v", resp)
}
actual := resp.Queries[0]
if resp.Index != actual.ModifyIndex {
t.Fatalf("bad index: %d", resp.Index)
}
actual.CreateIndex, actual.ModifyIndex = 0, 0
if !reflect.DeepEqual(actual, query.Query) {
t.Fatalf("bad: %v", actual)
}
}
}
func TestPreparedQuery_Explain(t *testing.T) {
t.Parallel()
2016-03-03 09:04:12 +00:00
dir1, s1 := testServerWithConfig(t, func(c *Config) {
c.ACLDatacenter = "dc1"
New ACLs (#4791) This PR is almost a complete rewrite of the ACL system within Consul. It brings the features more in line with other HashiCorp products. Obviously there is quite a bit left to do here but most of it is related docs, testing and finishing the last few commands in the CLI. I will update the PR description and check off the todos as I finish them over the next few days/week. Description At a high level this PR is mainly to split ACL tokens from Policies and to split the concepts of Authorization from Identities. A lot of this PR is mostly just to support CRUD operations on ACLTokens and ACLPolicies. These in and of themselves are not particularly interesting. The bigger conceptual changes are in how tokens get resolved, how backwards compatibility is handled and the separation of policy from identity which could lead the way to allowing for alternative identity providers. On the surface and with a new cluster the ACL system will look very similar to that of Nomads. Both have tokens and policies. Both have local tokens. The ACL management APIs for both are very similar. I even ripped off Nomad's ACL bootstrap resetting procedure. There are a few key differences though. Nomad requires token and policy replication where Consul only requires policy replication with token replication being opt-in. In Consul local tokens only work with token replication being enabled though. All policies in Nomad are globally applicable. In Consul all policies are stored and replicated globally but can be scoped to a subset of the datacenters. This allows for more granular access management. Unlike Nomad, Consul has legacy baggage in the form of the original ACL system. The ramifications of this are: A server running the new system must still support other clients using the legacy system. A client running the new system must be able to use the legacy RPCs when the servers in its datacenter are running the legacy system. The primary ACL DC's servers running in legacy mode needs to be a gate that keeps everything else in the entire multi-DC cluster running in legacy mode. So not only does this PR implement the new ACL system but has a legacy mode built in for when the cluster isn't ready for new ACLs. Also detecting that new ACLs can be used is automatic and requires no configuration on the part of administrators. This process is detailed more in the "Transitioning from Legacy to New ACL Mode" section below.
2018-10-19 16:04:07 +00:00
c.ACLsEnabled = true
2016-03-03 09:04:12 +00:00
c.ACLMasterToken = "root"
c.ACLDefaultPolicy = "deny"
})
defer os.RemoveAll(dir1)
defer s1.Shutdown()
codec := rpcClient(t, s1)
defer codec.Close()
testrpc.WaitForLeader(t, s1.RPC, "dc1", testrpc.WithToken("root"))
2016-03-03 09:04:12 +00:00
// Create an ACL with write permissions for prod- queries.
var token string
{
var rules = `
query "prod-" {
policy = "write"
}
`
req := structs.ACLRequest{
Datacenter: "dc1",
Op: structs.ACLSet,
ACL: structs.ACL{
Name: "User token",
New ACLs (#4791) This PR is almost a complete rewrite of the ACL system within Consul. It brings the features more in line with other HashiCorp products. Obviously there is quite a bit left to do here but most of it is related docs, testing and finishing the last few commands in the CLI. I will update the PR description and check off the todos as I finish them over the next few days/week. Description At a high level this PR is mainly to split ACL tokens from Policies and to split the concepts of Authorization from Identities. A lot of this PR is mostly just to support CRUD operations on ACLTokens and ACLPolicies. These in and of themselves are not particularly interesting. The bigger conceptual changes are in how tokens get resolved, how backwards compatibility is handled and the separation of policy from identity which could lead the way to allowing for alternative identity providers. On the surface and with a new cluster the ACL system will look very similar to that of Nomads. Both have tokens and policies. Both have local tokens. The ACL management APIs for both are very similar. I even ripped off Nomad's ACL bootstrap resetting procedure. There are a few key differences though. Nomad requires token and policy replication where Consul only requires policy replication with token replication being opt-in. In Consul local tokens only work with token replication being enabled though. All policies in Nomad are globally applicable. In Consul all policies are stored and replicated globally but can be scoped to a subset of the datacenters. This allows for more granular access management. Unlike Nomad, Consul has legacy baggage in the form of the original ACL system. The ramifications of this are: A server running the new system must still support other clients using the legacy system. A client running the new system must be able to use the legacy RPCs when the servers in its datacenter are running the legacy system. The primary ACL DC's servers running in legacy mode needs to be a gate that keeps everything else in the entire multi-DC cluster running in legacy mode. So not only does this PR implement the new ACL system but has a legacy mode built in for when the cluster isn't ready for new ACLs. Also detecting that new ACLs can be used is automatic and requires no configuration on the part of administrators. This process is detailed more in the "Transitioning from Legacy to New ACL Mode" section below.
2018-10-19 16:04:07 +00:00
Type: structs.ACLTokenTypeClient,
2016-03-03 09:04:12 +00:00
Rules: rules,
},
WriteRequest: structs.WriteRequest{Token: "root"},
}
if err := msgpackrpc.CallWithCodec(codec, "ACL.Apply", &req, &token); err != nil {
t.Fatalf("err: %v", err)
}
}
// Set up a template.
query := structs.PreparedQueryRequest{
Datacenter: "dc1",
Op: structs.PreparedQueryCreate,
Query: &structs.PreparedQuery{
Name: "prod-",
Token: "5e1e24e5-1329-f86f-18c6-3d3734edb2cd",
Template: structs.QueryTemplateOptions{
Type: structs.QueryTemplateTypeNamePrefixMatch,
},
Service: structs.ServiceQuery{
Service: "${name.full}",
},
},
WriteRequest: structs.WriteRequest{Token: token},
}
var reply string
if err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply); err != nil {
t.Fatalf("err: %v", err)
}
// Explain via the management token.
2016-03-03 09:04:12 +00:00
query.Query.ID = reply
query.Query.Service.Service = "prod-redis"
{
req := &structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: "prod-redis",
QueryOptions: structs.QueryOptions{Token: "root"},
}
var resp structs.PreparedQueryExplainResponse
err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Explain", req, &resp)
2016-03-03 09:04:12 +00:00
if err != nil {
t.Fatalf("err: %v", err)
}
actual := &resp.Query
actual.CreateIndex, actual.ModifyIndex = 0, 0
if !reflect.DeepEqual(actual, query.Query) {
t.Fatalf("bad: %v", actual)
}
}
// Explain via the user token, which will redact the captured token.
2016-03-03 09:04:12 +00:00
query.Query.Token = redactedToken
query.Query.Service.Service = "prod-redis"
{
req := &structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: "prod-redis",
QueryOptions: structs.QueryOptions{Token: token},
}
var resp structs.PreparedQueryExplainResponse
err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Explain", req, &resp)
2016-03-03 09:04:12 +00:00
if err != nil {
t.Fatalf("err: %v", err)
}
actual := &resp.Query
actual.CreateIndex, actual.ModifyIndex = 0, 0
if !reflect.DeepEqual(actual, query.Query) {
t.Fatalf("bad: %v", actual)
}
}
// Explaining should be denied without a token, since the user isn't
2016-03-03 09:04:12 +00:00
// allowed to see the query.
{
req := &structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: "prod-redis",
}
var resp structs.PreparedQueryExplainResponse
err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Explain", req, &resp)
if !acl.IsErrPermissionDenied(err) {
2016-03-03 09:04:12 +00:00
t.Fatalf("bad: %v", err)
}
}
// Try to explain a bogus ID.
2016-03-03 09:04:12 +00:00
{
req := &structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: generateUUID(),
QueryOptions: structs.QueryOptions{Token: "root"},
}
var resp structs.IndexedPreparedQueries
if err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Explain", req, &resp); err != nil {
if !structs.IsErrQueryNotFound(err) {
2016-03-03 09:04:12 +00:00
t.Fatalf("err: %v", err)
}
}
}
}
// This is a beast of a test, but the setup is so extensive it makes sense to
// walk through the different cases once we have it up. This is broken into
// sections so it's still pretty easy to read.
func TestPreparedQuery_Execute(t *testing.T) {
t.Parallel()
dir1, s1 := testServerWithConfig(t, func(c *Config) {
c.ACLDatacenter = "dc1"
New ACLs (#4791) This PR is almost a complete rewrite of the ACL system within Consul. It brings the features more in line with other HashiCorp products. Obviously there is quite a bit left to do here but most of it is related docs, testing and finishing the last few commands in the CLI. I will update the PR description and check off the todos as I finish them over the next few days/week. Description At a high level this PR is mainly to split ACL tokens from Policies and to split the concepts of Authorization from Identities. A lot of this PR is mostly just to support CRUD operations on ACLTokens and ACLPolicies. These in and of themselves are not particularly interesting. The bigger conceptual changes are in how tokens get resolved, how backwards compatibility is handled and the separation of policy from identity which could lead the way to allowing for alternative identity providers. On the surface and with a new cluster the ACL system will look very similar to that of Nomads. Both have tokens and policies. Both have local tokens. The ACL management APIs for both are very similar. I even ripped off Nomad's ACL bootstrap resetting procedure. There are a few key differences though. Nomad requires token and policy replication where Consul only requires policy replication with token replication being opt-in. In Consul local tokens only work with token replication being enabled though. All policies in Nomad are globally applicable. In Consul all policies are stored and replicated globally but can be scoped to a subset of the datacenters. This allows for more granular access management. Unlike Nomad, Consul has legacy baggage in the form of the original ACL system. The ramifications of this are: A server running the new system must still support other clients using the legacy system. A client running the new system must be able to use the legacy RPCs when the servers in its datacenter are running the legacy system. The primary ACL DC's servers running in legacy mode needs to be a gate that keeps everything else in the entire multi-DC cluster running in legacy mode. So not only does this PR implement the new ACL system but has a legacy mode built in for when the cluster isn't ready for new ACLs. Also detecting that new ACLs can be used is automatic and requires no configuration on the part of administrators. This process is detailed more in the "Transitioning from Legacy to New ACL Mode" section below.
2018-10-19 16:04:07 +00:00
c.ACLsEnabled = true
c.ACLMasterToken = "root"
c.ACLDefaultPolicy = "deny"
})
defer os.RemoveAll(dir1)
defer s1.Shutdown()
waitForLeaderEstablishment(t, s1)
codec1 := rpcClient(t, s1)
defer codec1.Close()
dir2, s2 := testServerWithConfig(t, func(c *Config) {
c.Datacenter = "dc2"
c.ACLDatacenter = "dc1"
New ACLs (#4791) This PR is almost a complete rewrite of the ACL system within Consul. It brings the features more in line with other HashiCorp products. Obviously there is quite a bit left to do here but most of it is related docs, testing and finishing the last few commands in the CLI. I will update the PR description and check off the todos as I finish them over the next few days/week. Description At a high level this PR is mainly to split ACL tokens from Policies and to split the concepts of Authorization from Identities. A lot of this PR is mostly just to support CRUD operations on ACLTokens and ACLPolicies. These in and of themselves are not particularly interesting. The bigger conceptual changes are in how tokens get resolved, how backwards compatibility is handled and the separation of policy from identity which could lead the way to allowing for alternative identity providers. On the surface and with a new cluster the ACL system will look very similar to that of Nomads. Both have tokens and policies. Both have local tokens. The ACL management APIs for both are very similar. I even ripped off Nomad's ACL bootstrap resetting procedure. There are a few key differences though. Nomad requires token and policy replication where Consul only requires policy replication with token replication being opt-in. In Consul local tokens only work with token replication being enabled though. All policies in Nomad are globally applicable. In Consul all policies are stored and replicated globally but can be scoped to a subset of the datacenters. This allows for more granular access management. Unlike Nomad, Consul has legacy baggage in the form of the original ACL system. The ramifications of this are: A server running the new system must still support other clients using the legacy system. A client running the new system must be able to use the legacy RPCs when the servers in its datacenter are running the legacy system. The primary ACL DC's servers running in legacy mode needs to be a gate that keeps everything else in the entire multi-DC cluster running in legacy mode. So not only does this PR implement the new ACL system but has a legacy mode built in for when the cluster isn't ready for new ACLs. Also detecting that new ACLs can be used is automatic and requires no configuration on the part of administrators. This process is detailed more in the "Transitioning from Legacy to New ACL Mode" section below.
2018-10-19 16:04:07 +00:00
c.ACLsEnabled = true
c.ACLDefaultPolicy = "deny"
})
defer os.RemoveAll(dir2)
defer s2.Shutdown()
waitForLeaderEstablishment(t, s2)
codec2 := rpcClient(t, s2)
defer codec2.Close()
s2.tokens.UpdateReplicationToken("root", tokenStore.TokenSourceConfig)
// Try to WAN join.
joinWAN(t, s2, s1)
retry.Run(t, func(r *retry.R) {
if got, want := len(s1.WANMembers()), 2; got != want {
r.Fatalf("got %d WAN members want %d", got, want)
}
if got, want := len(s2.WANMembers()), 2; got != want {
r.Fatalf("got %d WAN members want %d", got, want)
}
})
// check for RPC forwarding
testrpc.WaitForLeader(t, s1.RPC, "dc1", testrpc.WithToken("root"))
testrpc.WaitForLeader(t, s1.RPC, "dc2", testrpc.WithToken("root"))
// Create ACL tokens with read permission to the service and to the service
// and all nodes.
var execNoNodesToken string
{
req := structs.ACLRequest{
Datacenter: "dc1",
Op: structs.ACLSet,
ACL: structs.ACL{
Name: "User token",
Type: structs.ACLTokenTypeClient,
Rules: `service "foo" { policy = "read" }`,
},
WriteRequest: structs.WriteRequest{Token: "root"},
}
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "ACL.Apply", &req, &execNoNodesToken))
}
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
var execToken string
{
req := structs.ACLRequest{
Datacenter: "dc1",
Op: structs.ACLSet,
ACL: structs.ACL{
Name: "User token",
Type: structs.ACLTokenTypeClient,
Rules: `service "foo" { policy = "read" }
node "" { policy = "read" }`,
},
WriteRequest: structs.WriteRequest{Token: "root"},
}
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "ACL.Apply", &req, &execToken))
}
// Make a new exec token that can't read the service.
var denyToken string
{
req := structs.ACLRequest{
Datacenter: "dc1",
Op: structs.ACLSet,
ACL: structs.ACL{
Name: "User token",
New ACLs (#4791) This PR is almost a complete rewrite of the ACL system within Consul. It brings the features more in line with other HashiCorp products. Obviously there is quite a bit left to do here but most of it is related docs, testing and finishing the last few commands in the CLI. I will update the PR description and check off the todos as I finish them over the next few days/week. Description At a high level this PR is mainly to split ACL tokens from Policies and to split the concepts of Authorization from Identities. A lot of this PR is mostly just to support CRUD operations on ACLTokens and ACLPolicies. These in and of themselves are not particularly interesting. The bigger conceptual changes are in how tokens get resolved, how backwards compatibility is handled and the separation of policy from identity which could lead the way to allowing for alternative identity providers. On the surface and with a new cluster the ACL system will look very similar to that of Nomads. Both have tokens and policies. Both have local tokens. The ACL management APIs for both are very similar. I even ripped off Nomad's ACL bootstrap resetting procedure. There are a few key differences though. Nomad requires token and policy replication where Consul only requires policy replication with token replication being opt-in. In Consul local tokens only work with token replication being enabled though. All policies in Nomad are globally applicable. In Consul all policies are stored and replicated globally but can be scoped to a subset of the datacenters. This allows for more granular access management. Unlike Nomad, Consul has legacy baggage in the form of the original ACL system. The ramifications of this are: A server running the new system must still support other clients using the legacy system. A client running the new system must be able to use the legacy RPCs when the servers in its datacenter are running the legacy system. The primary ACL DC's servers running in legacy mode needs to be a gate that keeps everything else in the entire multi-DC cluster running in legacy mode. So not only does this PR implement the new ACL system but has a legacy mode built in for when the cluster isn't ready for new ACLs. Also detecting that new ACLs can be used is automatic and requires no configuration on the part of administrators. This process is detailed more in the "Transitioning from Legacy to New ACL Mode" section below.
2018-10-19 16:04:07 +00:00
Type: structs.ACLTokenTypeClient,
Rules: `service "foo" { policy = "deny" }`,
},
WriteRequest: structs.WriteRequest{Token: "root"},
}
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "ACL.Apply", &req, &denyToken))
}
newSessionDC1 := func(t *testing.T) string {
t.Helper()
req := structs.SessionRequest{
Datacenter: "dc1",
Op: structs.SessionCreate,
Session: structs.Session{
Node: s1.config.NodeName,
},
WriteRequest: structs.WriteRequest{Token: "root"},
}
var session string
if err := msgpackrpc.CallWithCodec(codec1, "Session.Apply", &req, &session); err != nil {
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
t.Fatalf("err: %v", err)
}
return session
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
}
// Set up some nodes in each DC that host the service.
{
for i := 0; i < 10; i++ {
for _, dc := range []string{"dc1", "dc2"} {
req := structs.RegisterRequest{
Datacenter: dc,
Node: fmt.Sprintf("node%d", i+1),
Address: fmt.Sprintf("127.0.0.%d", i+1),
NodeMeta: map[string]string{
"group": fmt.Sprintf("%d", i/5),
"instance_type": "t2.micro",
},
Service: &structs.NodeService{
Service: "foo",
Port: 8000,
Tags: []string{dc, fmt.Sprintf("tag%d", i+1)},
Meta: map[string]string{
"svc-group": fmt.Sprintf("%d", i%2),
"foo": "true",
},
},
WriteRequest: structs.WriteRequest{Token: "root"},
}
if i == 0 {
req.NodeMeta["unique"] = "true"
req.Service.Meta["unique"] = "true"
}
var codec rpc.ClientCodec
if dc == "dc1" {
codec = codec1
} else {
codec = codec2
}
var reply struct{}
if err := msgpackrpc.CallWithCodec(codec, "Catalog.Register", &req, &reply); err != nil {
t.Fatalf("err: %v", err)
}
}
}
}
// Set up a service query.
query := structs.PreparedQueryRequest{
Datacenter: "dc1",
Op: structs.PreparedQueryCreate,
Query: &structs.PreparedQuery{
Name: "test",
Service: structs.ServiceQuery{
Service: "foo",
},
DNS: structs.QueryDNSOptions{
TTL: "10s",
},
},
WriteRequest: structs.WriteRequest{Token: "root"},
}
if err := msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Apply", &query, &query.Query.ID); err != nil {
t.Fatalf("err: %v", err)
}
// Run a query that doesn't exist.
t.Run("run query that doesn't exist", func(t *testing.T) {
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: "nope",
}
var reply structs.PreparedQueryExecuteResponse
err := msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply)
assert.EqualError(t, err, structs.ErrQueryNotFound.Error())
assert.Len(t, reply.Nodes, 0)
})
expectNodes := func(t *testing.T, query *structs.PreparedQueryRequest, reply *structs.PreparedQueryExecuteResponse, n int) {
t.Helper()
assert.Len(t, reply.Nodes, n)
assert.Equal(t, "dc1", reply.Datacenter)
assert.Equal(t, 0, reply.Failovers)
assert.Equal(t, query.Query.Service.Service, reply.Service)
assert.Equal(t, query.Query.DNS, reply.DNS)
assert.True(t, reply.QueryMeta.KnownLeader)
}
expectFailoverNodes := func(t *testing.T, query *structs.PreparedQueryRequest, reply *structs.PreparedQueryExecuteResponse, n int) {
t.Helper()
assert.Len(t, reply.Nodes, n)
assert.Equal(t, "dc2", reply.Datacenter)
assert.Equal(t, 1, reply.Failovers)
assert.Equal(t, query.Query.Service.Service, reply.Service)
assert.Equal(t, query.Query.DNS, reply.DNS)
assert.True(t, reply.QueryMeta.KnownLeader)
}
t.Run("run the registered query", func(t *testing.T) {
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
QueryOptions: structs.QueryOptions{Token: execToken},
}
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
expectNodes(t, &query, &reply, 10)
})
t.Run("try with a limit", func(t *testing.T) {
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
Limit: 3,
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
QueryOptions: structs.QueryOptions{Token: execToken},
}
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
expectNodes(t, &query, &reply, 3)
})
// Run various service queries with node metadata filters.
for name, tc := range map[string]struct {
filters map[string]string
numNodes int
}{
"no filter 10 nodes": {
filters: map[string]string{},
numNodes: 10,
},
"instance filter 10 nodes": {
filters: map[string]string{"instance_type": "t2.micro"},
numNodes: 10,
},
"group filter 5 nodes": {
filters: map[string]string{"group": "1"},
numNodes: 5,
},
"group filter unique 1 node": {
filters: map[string]string{"group": "0", "unique": "true"},
numNodes: 1,
},
} {
tc := tc
t.Run("node metadata - "+name, func(t *testing.T) {
session := newSessionDC1(t)
nodeMetaQuery := structs.PreparedQueryRequest{
Datacenter: "dc1",
Op: structs.PreparedQueryCreate,
Query: &structs.PreparedQuery{
Session: session,
Service: structs.ServiceQuery{
Service: "foo",
NodeMeta: tc.filters,
},
DNS: structs.QueryDNSOptions{
TTL: "10s",
},
},
WriteRequest: structs.WriteRequest{Token: "root"},
}
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Apply", &nodeMetaQuery, &nodeMetaQuery.Query.ID))
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: nodeMetaQuery.Query.ID,
QueryOptions: structs.QueryOptions{Token: execToken},
}
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
assert.Len(t, reply.Nodes, tc.numNodes)
for _, node := range reply.Nodes {
assert.True(t, structs.SatisfiesMetaFilters(node.Node.Meta, tc.filters), "meta: %v", node.Node.Meta)
}
})
}
// Run various service queries with service metadata filters
for name, tc := range map[string]struct {
filters map[string]string
numNodes int
}{
"no filter 10 nodes": {
filters: map[string]string{},
numNodes: 10,
},
"foo filter 10 nodes": {
filters: map[string]string{"foo": "true"},
numNodes: 10,
},
"group filter 0 - 5 nodes": {
filters: map[string]string{"svc-group": "0"},
numNodes: 5,
},
"group filter 1 - 5 nodes": {
filters: map[string]string{"svc-group": "1"},
numNodes: 5,
},
"group filter 0 - unique 1 node": {
filters: map[string]string{"svc-group": "0", "unique": "true"},
numNodes: 1,
},
} {
tc := tc
require.True(t, t.Run("service metadata - "+name, func(t *testing.T) {
session := newSessionDC1(t)
svcMetaQuery := structs.PreparedQueryRequest{
Datacenter: "dc1",
Op: structs.PreparedQueryCreate,
Query: &structs.PreparedQuery{
Session: session,
Service: structs.ServiceQuery{
Service: "foo",
ServiceMeta: tc.filters,
},
DNS: structs.QueryDNSOptions{
TTL: "10s",
},
},
WriteRequest: structs.WriteRequest{Token: "root"},
}
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Apply", &svcMetaQuery, &svcMetaQuery.Query.ID))
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: svcMetaQuery.Query.ID,
QueryOptions: structs.QueryOptions{Token: execToken},
}
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
assert.Len(t, reply.Nodes, tc.numNodes)
for _, node := range reply.Nodes {
assert.True(t, structs.SatisfiesMetaFilters(node.Service.Meta, tc.filters), "meta: %v", node.Service.Meta)
}
}))
}
// Push a coordinate for one of the nodes so we can try an RTT sort. We
// have to sleep a little while for the coordinate batch to get flushed.
{
req := structs.CoordinateUpdateRequest{
Datacenter: "dc1",
Node: "node3",
Coord: coordinate.NewCoordinate(coordinate.DefaultConfig()),
WriteRequest: structs.WriteRequest{Token: "root"},
}
var out struct{}
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "Coordinate.Update", &req, &out))
time.Sleep(3 * s1.config.CoordinateUpdatePeriod)
}
// Try an RTT sort. We don't have any other coordinates in there but
// showing that the node with a coordinate is always first proves we
// call the RTT sorting function, which is tested elsewhere.
for i := 0; i < 100; i++ {
t.Run(fmt.Sprintf("rtt sort iter %d", i), func(t *testing.T) {
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
Source: structs.QuerySource{
Datacenter: "dc1",
Node: "node3",
},
QueryOptions: structs.QueryOptions{Token: execToken},
}
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
expectNodes(t, &query, &reply, 10)
assert.Equal(t, "node3", reply.Nodes[0].Node.Node)
})
}
// Make sure the shuffle looks like it's working.
uniques := make(map[string]struct{})
for i := 0; i < 100; i++ {
t.Run(fmt.Sprintf("shuffle iter %d", i), func(t *testing.T) {
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: execToken},
}
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
expectNodes(t, &query, &reply, 10)
var names []string
for _, node := range reply.Nodes {
names = append(names, node.Node.Node)
}
key := strings.Join(names, "|")
uniques[key] = struct{}{}
})
}
// We have to allow for the fact that there won't always be a unique
// shuffle each pass, so we just look for smell here without the test
// being flaky.
if len(uniques) < 50 {
t.Fatalf("unique shuffle ratio too low: %d/100", len(uniques))
}
// Set the query to return results nearest to node3. This is the only
// node with coordinates, and it carries the service we are asking for,
// so node3 should always show up first.
query.Op = structs.PreparedQueryUpdate
query.Query.Service.Near = "node3"
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Apply", &query, &query.Query.ID))
// Now run the query and make sure the sort looks right.
for i := 0; i < 10; i++ {
t.Run(fmt.Sprintf("run nearest query iter %d", i), func(t *testing.T) {
req := structs.PreparedQueryExecuteRequest{
Agent: structs.QuerySource{
Datacenter: "dc1",
Node: "node3",
},
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: execToken},
}
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
assert.Len(t, reply.Nodes, 10)
assert.Equal(t, "node3", reply.Nodes[0].Node.Node)
})
}
// Query again, but this time set a client-supplied query source. This
// proves that we allow overriding the baked-in value with ?near.
t.Run("nearest fallback to shuffle", func(t *testing.T) {
// Set up the query with a non-existent node. This will cause the
// nodes to be shuffled if the passed node is respected, proving
// that we allow the override to happen.
req := structs.PreparedQueryExecuteRequest{
Source: structs.QuerySource{
Datacenter: "dc1",
Node: "foo",
},
Agent: structs.QuerySource{
Datacenter: "dc1",
Node: "node3",
},
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: execToken},
}
shuffled := false
for i := 0; i < 10; i++ {
2017-01-24 02:11:13 +00:00
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
assert.Len(t, reply.Nodes, 10)
if node := reply.Nodes[0].Node.Node; node != "node3" {
shuffled = true
break
}
}
require.True(t, shuffled, "expect nodes to be shuffled")
})
// If the exact node we are sorting near appears in the list, make sure it
// gets popped to the front of the result.
t.Run("nearest bypasses shuffle", func(t *testing.T) {
req := structs.PreparedQueryExecuteRequest{
Source: structs.QuerySource{
Datacenter: "dc1",
Node: "node1",
},
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: execToken},
}
for i := 0; i < 10; i++ {
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
assert.Len(t, reply.Nodes, 10)
assert.Equal(t, "node1", reply.Nodes[0].Node.Node)
}
})
// Bake the magic "_agent" flag into the query.
query.Query.Service.Near = "_agent"
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Apply", &query, &query.Query.ID))
// Check that we sort the local agent first when the magic flag is set.
t.Run("local agent is first using _agent on node3", func(t *testing.T) {
req := structs.PreparedQueryExecuteRequest{
Agent: structs.QuerySource{
Datacenter: "dc1",
Node: "node3",
},
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: execToken},
}
for i := 0; i < 10; i++ {
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
assert.Len(t, reply.Nodes, 10)
assert.Equal(t, "node3", reply.Nodes[0].Node.Node)
}
})
2016-06-21 19:54:18 +00:00
// Check that the query isn't just sorting "node3" first because we
// provided it in the Agent query source. Proves that we use the
// Agent source when the magic "_agent" flag is passed.
t.Run("local agent is first using _agent on foo", func(t *testing.T) {
2016-06-21 19:54:18 +00:00
req := structs.PreparedQueryExecuteRequest{
Agent: structs.QuerySource{
Datacenter: "dc1",
Node: "foo",
},
2016-06-21 19:54:18 +00:00
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: execToken},
}
// Expect the set to be shuffled since we have no coordinates
// on the "foo" node.
shuffled := false
2016-06-21 19:54:18 +00:00
for i := 0; i < 10; i++ {
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
assert.Len(t, reply.Nodes, 10)
2016-06-21 19:54:18 +00:00
if node := reply.Nodes[0].Node.Node; node != "node3" {
shuffled = true
break
}
}
require.True(t, shuffled, "expect nodes to be shuffled")
})
// Shuffles if the response comes from a non-local DC. Proves that the
// agent query source does not interfere with the order.
t.Run("shuffles if coming from non-local dc", func(t *testing.T) {
req := structs.PreparedQueryExecuteRequest{
Source: structs.QuerySource{
Datacenter: "dc2",
Node: "node3",
},
Agent: structs.QuerySource{
Datacenter: "dc1",
Node: "node3",
},
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: execToken},
}
shuffled := false
for i := 0; i < 10; i++ {
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
assert.Len(t, reply.Nodes, 10)
if reply.Nodes[0].Node.Node != "node3" {
shuffled = true
break
}
}
require.True(t, shuffled, "expect node shuffle for remote results")
})
2016-06-21 19:54:18 +00:00
// Un-bake the near parameter.
query.Query.Service.Near = ""
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Apply", &query, &query.Query.ID))
// Update the health of a node to mark it critical.
setHealth := func(t *testing.T, node string, health string) {
t.Helper()
req := structs.RegisterRequest{
Datacenter: "dc1",
Node: node,
Address: "127.0.0.1",
Service: &structs.NodeService{
Service: "foo",
Port: 8000,
Tags: []string{"dc1", "tag1"},
},
Check: &structs.HealthCheck{
Name: "failing",
Status: health,
ServiceID: "foo",
},
WriteRequest: structs.WriteRequest{Token: "root"},
}
var reply struct{}
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "Catalog.Register", &req, &reply))
}
setHealth(t, "node1", api.HealthCritical)
// The failing node should be filtered.
t.Run("failing node filtered", func(t *testing.T) {
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
QueryOptions: structs.QueryOptions{Token: execToken},
}
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
expectNodes(t, &query, &reply, 9)
for _, node := range reply.Nodes {
assert.NotEqual(t, "node1", node.Node.Node)
}
})
// Upgrade it to a warning and re-query, should be 10 nodes again.
setHealth(t, "node1", api.HealthWarning)
t.Run("warning nodes are included", func(t *testing.T) {
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
QueryOptions: structs.QueryOptions{Token: execToken},
}
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
expectNodes(t, &query, &reply, 10)
})
// Make the query more picky so it excludes warning nodes.
query.Query.Service.OnlyPassing = true
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Apply", &query, &query.Query.ID))
// The node in the warning state should be filtered.
t.Run("warning nodes are omitted with onlypassing=true", func(t *testing.T) {
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
QueryOptions: structs.QueryOptions{Token: execToken},
}
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
expectNodes(t, &query, &reply, 9)
for _, node := range reply.Nodes {
assert.NotEqual(t, "node1", node.Node.Node)
}
})
// Make the query ignore all our health checks (which have "failing" ID
// implicitly from their name).
query.Query.Service.IgnoreCheckIDs = []types.CheckID{"failing"}
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Apply", &query, &query.Query.ID))
// We should end up with 10 nodes again
t.Run("all nodes including when ignoring failing checks", func(t *testing.T) {
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: execToken},
}
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
expectNodes(t, &query, &reply, 10)
})
// Undo that so all the following tests aren't broken!
query.Query.Service.IgnoreCheckIDs = nil
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Apply", &query, &query.Query.ID))
// Make the query more picky by adding a tag filter. This just proves we
// call into the tag filter, it is tested more thoroughly in a separate
// test.
query.Query.Service.Tags = []string{"!tag3"}
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Apply", &query, &query.Query.ID))
// The node in the warning state should be filtered as well as the node
// with the filtered tag.
t.Run("filter node in warning state and filtered node", func(t *testing.T) {
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
QueryOptions: structs.QueryOptions{Token: execToken},
}
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
expectNodes(t, &query, &reply, 8)
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
for _, node := range reply.Nodes {
assert.NotEqual(t, "node1", node.Node.Node)
assert.NotEqual(t, "node3", node.Node.Node)
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
}
})
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
// Make sure the query gets denied with this token.
t.Run("query denied with deny token", func(t *testing.T) {
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: denyToken},
}
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
expectNodes(t, &query, &reply, 0)
})
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
// Bake the exec token into the query.
query.Query.Token = execToken
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Apply", &query, &query.Query.ID))
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
// Now even querying with the deny token should work.
t.Run("query with deny token still works", func(t *testing.T) {
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: denyToken},
}
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
expectNodes(t, &query, &reply, 8)
for _, node := range reply.Nodes {
assert.NotEqual(t, "node1", node.Node.Node)
assert.NotEqual(t, "node3", node.Node.Node)
}
})
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
// Un-bake the token.
query.Query.Token = ""
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Apply", &query, &query.Query.ID))
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
// Make sure the query gets denied again with the deny token.
t.Run("denied with deny token when no query token", func(t *testing.T) {
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: denyToken},
}
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
expectNodes(t, &query, &reply, 0)
})
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
t.Run("filter nodes with exec token without node privileges", func(t *testing.T) {
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: execNoNodesToken},
}
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
expectNodes(t, &query, &reply, 0)
})
t.Run("normal operation again with exec token", func(t *testing.T) {
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: execToken},
}
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
expectNodes(t, &query, &reply, 8)
for _, node := range reply.Nodes {
assert.NotEqual(t, "node1", node.Node.Node)
assert.NotEqual(t, "node3", node.Node.Node)
}
})
// Now fail everything in dc1 and we should get an empty list back.
for i := 0; i < 10; i++ {
setHealth(t, fmt.Sprintf("node%d", i+1), api.HealthCritical)
}
t.Run("everything is failing so should get empty list", func(t *testing.T) {
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
QueryOptions: structs.QueryOptions{Token: execToken},
}
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
expectNodes(t, &query, &reply, 0)
})
// Modify the query to have it fail over to a bogus DC and then dc2.
query.Query.Service.Failover.Datacenters = []string{"bogus", "dc2"}
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Apply", &query, &query.Query.ID))
// Now we should see 9 nodes from dc2 (we have the tag filter still).
t.Run("see 9 nodes from dc2 using tag filter", func(t *testing.T) {
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
QueryOptions: structs.QueryOptions{Token: execToken},
}
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
expectFailoverNodes(t, &query, &reply, 9)
for _, node := range reply.Nodes {
assert.NotEqual(t, "node3", node.Node.Node)
}
})
// Make sure the limit and query options are forwarded.
t.Run("forward limit and query options", func(t *testing.T) {
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
Limit: 3,
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
QueryOptions: structs.QueryOptions{
Token: execToken,
RequireConsistent: true,
},
}
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
expectFailoverNodes(t, &query, &reply, 3)
for _, node := range reply.Nodes {
assert.NotEqual(t, "node3", node.Node.Node)
}
})
// Make sure the remote shuffle looks like it's working.
uniques = make(map[string]struct{})
for i := 0; i < 100; i++ {
t.Run(fmt.Sprintf("remote shuffle iter %d", i), func(t *testing.T) {
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: execToken},
}
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
expectFailoverNodes(t, &query, &reply, 9)
var names []string
for _, node := range reply.Nodes {
names = append(names, node.Node.Node)
}
key := strings.Join(names, "|")
uniques[key] = struct{}{}
})
}
// We have to allow for the fact that there won't always be a unique
// shuffle each pass, so we just look for smell here without the test
// being flaky.
if len(uniques) < 50 {
t.Fatalf("unique shuffle ratio too low: %d/100", len(uniques))
}
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
// Make sure the query response from dc2 gets denied with the deny token.
t.Run("query from dc2 denied with deny token", func(t *testing.T) {
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
QueryOptions: structs.QueryOptions{Token: denyToken},
}
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
expectFailoverNodes(t, &query, &reply, 0)
})
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
// Bake the exec token into the query.
query.Query.Token = execToken
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Apply", &query, &query.Query.ID))
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
// Now even querying with the deny token should work.
t.Run("query from dc2 with exec token using deny token works", func(t *testing.T) {
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
QueryOptions: structs.QueryOptions{Token: denyToken},
}
var reply structs.PreparedQueryExecuteResponse
require.NoError(t, msgpackrpc.CallWithCodec(codec1, "PreparedQuery.Execute", &req, &reply))
expectFailoverNodes(t, &query, &reply, 9)
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
for _, node := range reply.Nodes {
assert.NotEqual(t, "node3", node.Node.Node)
Creates new "prepared-query" ACL type and new token capture behavior. Prior to this change, prepared queries had the following behavior for ACLs, which will need to change to support templates: 1. A management token, or a token with read access to the service being queried needed to be provided in order to create a prepared query. 2. The token used to create the prepared query was stored with the query in the state store and used to execute the query. 3. A management token, or the token used to create the query needed to be supplied to perform and CRUD operations on an existing prepared query. This was pretty subtle and complicated behavior, and won't work for templates since the service name is computed at execution time. To solve this, we introduce a new "prepared-query" ACL type, where the prefix applies to the query name for static prepared query types and to the prefix for template prepared query types. With this change, the new behavior is: 1. A management token, or a token with "prepared-query" write access to the query name or (soon) the given template prefix is required to do any CRUD operations on a prepared query, or to list prepared queries (the list is filtered by this ACL). 2. You will no longer need a management token to list prepared queries, but you will only be able to see prepared queries that you have access to (you get an empty list instead of permission denied). 3. When listing or getting a query, because it was easy to capture management tokens given the past behavior, this will always blank out the "Token" field (replacing the contents as <hidden>) for all tokens unless a management token is supplied. Going forward, we should discourage people from binding tokens for execution unless strictly necessary. 4. No token will be captured by default when a prepared query is created. If the user wishes to supply an execution token then can pass it in via the "Token" field in the prepared query definition. Otherwise, this field will default to empty. 5. At execution time, we will use the captured token if it exists with the prepared query definition, otherwise we will use the token that's passed in with the request, just like we do for other RPCs (or you can use the agent's configured token for DNS). 6. Prepared queries with no name (accessible only by ID) will not require ACLs to create or modify (execution time will depend on the service ACL configuration). Our argument here is that these are designed to be ephemeral and the IDs are as good as an ACL. Management tokens will be able to list all of these. These changes enable templates, but also enable delegation of authority to manage the prepared query namespace.
2016-02-23 08:12:58 +00:00
}
})
}
func TestPreparedQuery_Execute_ForwardLeader(t *testing.T) {
t.Parallel()
dir1, s1 := testServer(t)
defer os.RemoveAll(dir1)
defer s1.Shutdown()
codec1 := rpcClient(t, s1)
defer codec1.Close()
dir2, s2 := testServer(t)
defer os.RemoveAll(dir2)
defer s2.Shutdown()
codec2 := rpcClient(t, s2)
defer codec2.Close()
// Try to join.
2017-05-05 10:29:49 +00:00
joinLAN(t, s2, s1)
testrpc.WaitForLeader(t, s1.RPC, "dc1")
testrpc.WaitForLeader(t, s2.RPC, "dc1")
// Use the follower as the client.
var codec rpc.ClientCodec
if !s1.IsLeader() {
codec = codec1
} else {
codec = codec2
}
// Set up a node and service in the catalog.
{
req := structs.RegisterRequest{
Datacenter: "dc1",
Node: "foo",
Address: "127.0.0.1",
Service: &structs.NodeService{
Service: "redis",
Tags: []string{"master"},
Port: 8000,
},
}
var reply struct{}
if err := msgpackrpc.CallWithCodec(codec, "Catalog.Register", &req, &reply); err != nil {
t.Fatalf("err: %v", err)
}
}
// Set up a bare bones query.
query := structs.PreparedQueryRequest{
Datacenter: "dc1",
Op: structs.PreparedQueryCreate,
Query: &structs.PreparedQuery{
Name: "test",
Service: structs.ServiceQuery{
Service: "redis",
},
},
}
var reply string
if err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Apply", &query, &reply); err != nil {
t.Fatalf("err: %v", err)
}
// Execute it through the follower.
{
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: reply,
}
var reply structs.PreparedQueryExecuteResponse
if err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Execute", &req, &reply); err != nil {
t.Fatalf("err: %v", err)
}
if len(reply.Nodes) != 1 {
t.Fatalf("bad: %v", reply)
}
}
// Execute it through the follower with consistency turned on.
{
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: reply,
QueryOptions: structs.QueryOptions{RequireConsistent: true},
}
var reply structs.PreparedQueryExecuteResponse
if err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.Execute", &req, &reply); err != nil {
t.Fatalf("err: %v", err)
}
if len(reply.Nodes) != 1 {
t.Fatalf("bad: %v", reply)
}
}
// Remote execute it through the follower.
{
req := structs.PreparedQueryExecuteRemoteRequest{
Datacenter: "dc1",
Query: *query.Query,
}
var reply structs.PreparedQueryExecuteResponse
if err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.ExecuteRemote", &req, &reply); err != nil {
t.Fatalf("err: %v", err)
}
if len(reply.Nodes) != 1 {
t.Fatalf("bad: %v", reply)
}
}
// Remote execute it through the follower with consistency turned on.
{
req := structs.PreparedQueryExecuteRemoteRequest{
Datacenter: "dc1",
Query: *query.Query,
QueryOptions: structs.QueryOptions{RequireConsistent: true},
}
var reply structs.PreparedQueryExecuteResponse
if err := msgpackrpc.CallWithCodec(codec, "PreparedQuery.ExecuteRemote", &req, &reply); err != nil {
t.Fatalf("err: %v", err)
}
if len(reply.Nodes) != 1 {
t.Fatalf("bad: %v", reply)
}
}
}
2015-11-11 02:23:37 +00:00
func TestPreparedQuery_Execute_ConnectExact(t *testing.T) {
t.Parallel()
require := require.New(t)
dir1, s1 := testServer(t)
defer os.RemoveAll(dir1)
defer s1.Shutdown()
codec := rpcClient(t, s1)
defer codec.Close()
// Setup 3 services on 3 nodes: one is non-Connect, one is Connect native,
// and one is a proxy to the non-Connect one.
for i := 0; i < 3; i++ {
req := structs.RegisterRequest{
Datacenter: "dc1",
Node: fmt.Sprintf("node%d", i+1),
Address: fmt.Sprintf("127.0.0.%d", i+1),
Service: &structs.NodeService{
Service: "foo",
Port: 8000,
},
}
switch i {
case 0:
// Default do nothing
case 1:
// Connect native
req.Service.Connect.Native = true
case 2:
// Connect proxy
req.Service.Kind = structs.ServiceKindConnectProxy
Add Proxy Upstreams to Service Definition (#4639) * Refactor Service Definition ProxyDestination. This includes: - Refactoring all internal structs used - Updated tests for both deprecated and new input for: - Agent Services endpoint response - Agent Service endpoint response - Agent Register endpoint - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Register - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Services endpoint response - Catalog Node endpoint response - Catalog Service endpoint response - Updated API tests for all of the above too (both deprecated and new forms of register) TODO: - config package changes for on-disk service definitions - proxy config endpoint - built-in proxy support for new fields * Agent proxy config endpoint updated with upstreams * Config file changes for upstreams. * Add upstream opaque config and update all tests to ensure it works everywhere. * Built in proxy working with new Upstreams config * Command fixes and deprecations * Fix key translation, upstream type defaults and a spate of other subtele bugs found with ned to end test scripts... TODO: tests still failing on one case that needs a fix. I think it's key translation for upstreams nested in Managed proxy struct. * Fix translated keys in API registration. ≈ * Fixes from docs - omit some empty undocumented fields in API - Bring back ServiceProxyDestination in Catalog responses to not break backwards compat - this was removed assuming it was only used internally. * Documentation updates for Upstreams in service definition * Fixes for tests broken by many refactors. * Enable travis on f-connect branch in this branch too. * Add consistent Deprecation comments to ProxyDestination uses * Update version number on deprecation notices, and correct upstream datacenter field with explanation in docs
2018-09-12 16:07:47 +00:00
req.Service.Proxy.DestinationServiceName = req.Service.Service
req.Service.Service = "proxy"
}
var reply struct{}
require.NoError(msgpackrpc.CallWithCodec(codec, "Catalog.Register", &req, &reply))
}
// The query, start with connect disabled
query := structs.PreparedQueryRequest{
Datacenter: "dc1",
Op: structs.PreparedQueryCreate,
Query: &structs.PreparedQuery{
Name: "test",
Service: structs.ServiceQuery{
Service: "foo",
},
DNS: structs.QueryDNSOptions{
TTL: "10s",
},
},
}
require.NoError(msgpackrpc.CallWithCodec(
codec, "PreparedQuery.Apply", &query, &query.Query.ID))
// In the future we'll run updates
query.Op = structs.PreparedQueryUpdate
// Run the registered query.
{
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
}
var reply structs.PreparedQueryExecuteResponse
require.NoError(msgpackrpc.CallWithCodec(
codec, "PreparedQuery.Execute", &req, &reply))
// Result should have two because it omits the proxy whose name
// doesn't match the query.
require.Len(reply.Nodes, 2)
require.Equal(query.Query.Service.Service, reply.Service)
require.Equal(query.Query.DNS, reply.DNS)
require.True(reply.QueryMeta.KnownLeader, "queried leader")
}
// Run with the Connect setting specified on the request
{
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
Connect: true,
}
var reply structs.PreparedQueryExecuteResponse
require.NoError(msgpackrpc.CallWithCodec(
codec, "PreparedQuery.Execute", &req, &reply))
// Result should have two because we should get the native AND
// the proxy (since the destination matches our service name).
require.Len(reply.Nodes, 2)
require.Equal(query.Query.Service.Service, reply.Service)
require.Equal(query.Query.DNS, reply.DNS)
require.True(reply.QueryMeta.KnownLeader, "queried leader")
// Make sure the native is the first one
if !reply.Nodes[0].Service.Connect.Native {
reply.Nodes[0], reply.Nodes[1] = reply.Nodes[1], reply.Nodes[0]
}
require.True(reply.Nodes[0].Service.Connect.Native, "native")
require.Equal(reply.Service, reply.Nodes[0].Service.Service)
require.Equal(structs.ServiceKindConnectProxy, reply.Nodes[1].Service.Kind)
Add Proxy Upstreams to Service Definition (#4639) * Refactor Service Definition ProxyDestination. This includes: - Refactoring all internal structs used - Updated tests for both deprecated and new input for: - Agent Services endpoint response - Agent Service endpoint response - Agent Register endpoint - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Register - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Services endpoint response - Catalog Node endpoint response - Catalog Service endpoint response - Updated API tests for all of the above too (both deprecated and new forms of register) TODO: - config package changes for on-disk service definitions - proxy config endpoint - built-in proxy support for new fields * Agent proxy config endpoint updated with upstreams * Config file changes for upstreams. * Add upstream opaque config and update all tests to ensure it works everywhere. * Built in proxy working with new Upstreams config * Command fixes and deprecations * Fix key translation, upstream type defaults and a spate of other subtele bugs found with ned to end test scripts... TODO: tests still failing on one case that needs a fix. I think it's key translation for upstreams nested in Managed proxy struct. * Fix translated keys in API registration. ≈ * Fixes from docs - omit some empty undocumented fields in API - Bring back ServiceProxyDestination in Catalog responses to not break backwards compat - this was removed assuming it was only used internally. * Documentation updates for Upstreams in service definition * Fixes for tests broken by many refactors. * Enable travis on f-connect branch in this branch too. * Add consistent Deprecation comments to ProxyDestination uses * Update version number on deprecation notices, and correct upstream datacenter field with explanation in docs
2018-09-12 16:07:47 +00:00
require.Equal(reply.Service, reply.Nodes[1].Service.Proxy.DestinationServiceName)
}
// Update the query
query.Query.Service.Connect = true
require.NoError(msgpackrpc.CallWithCodec(
codec, "PreparedQuery.Apply", &query, &query.Query.ID))
// Run the registered query.
{
req := structs.PreparedQueryExecuteRequest{
Datacenter: "dc1",
QueryIDOrName: query.Query.ID,
}
var reply structs.PreparedQueryExecuteResponse
require.NoError(msgpackrpc.CallWithCodec(
codec, "PreparedQuery.Execute", &req, &reply))
// Result should have two because we should get the native AND
// the proxy (since the destination matches our service name).
require.Len(reply.Nodes, 2)
require.Equal(query.Query.Service.Service, reply.Service)
require.Equal(query.Query.DNS, reply.DNS)
require.True(reply.QueryMeta.KnownLeader, "queried leader")
// Make sure the native is the first one
if !reply.Nodes[0].Service.Connect.Native {
reply.Nodes[0], reply.Nodes[1] = reply.Nodes[1], reply.Nodes[0]
}
require.True(reply.Nodes[0].Service.Connect.Native, "native")
require.Equal(reply.Service, reply.Nodes[0].Service.Service)
require.Equal(structs.ServiceKindConnectProxy, reply.Nodes[1].Service.Kind)
Add Proxy Upstreams to Service Definition (#4639) * Refactor Service Definition ProxyDestination. This includes: - Refactoring all internal structs used - Updated tests for both deprecated and new input for: - Agent Services endpoint response - Agent Service endpoint response - Agent Register endpoint - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Register - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Services endpoint response - Catalog Node endpoint response - Catalog Service endpoint response - Updated API tests for all of the above too (both deprecated and new forms of register) TODO: - config package changes for on-disk service definitions - proxy config endpoint - built-in proxy support for new fields * Agent proxy config endpoint updated with upstreams * Config file changes for upstreams. * Add upstream opaque config and update all tests to ensure it works everywhere. * Built in proxy working with new Upstreams config * Command fixes and deprecations * Fix key translation, upstream type defaults and a spate of other subtele bugs found with ned to end test scripts... TODO: tests still failing on one case that needs a fix. I think it's key translation for upstreams nested in Managed proxy struct. * Fix translated keys in API registration. ≈ * Fixes from docs - omit some empty undocumented fields in API - Bring back ServiceProxyDestination in Catalog responses to not break backwards compat - this was removed assuming it was only used internally. * Documentation updates for Upstreams in service definition * Fixes for tests broken by many refactors. * Enable travis on f-connect branch in this branch too. * Add consistent Deprecation comments to ProxyDestination uses * Update version number on deprecation notices, and correct upstream datacenter field with explanation in docs
2018-09-12 16:07:47 +00:00
require.Equal(reply.Service, reply.Nodes[1].Service.Proxy.DestinationServiceName)
}
// Unset the query
query.Query.Service.Connect = false
require.NoError(msgpackrpc.CallWithCodec(
codec, "PreparedQuery.Apply", &query, &query.Query.ID))
}
2015-11-11 02:23:37 +00:00
func TestPreparedQuery_tagFilter(t *testing.T) {
t.Parallel()
2015-11-11 02:23:37 +00:00
testNodes := func() structs.CheckServiceNodes {
return structs.CheckServiceNodes{
structs.CheckServiceNode{
Node: &structs.Node{Node: "node1"},
Service: &structs.NodeService{Tags: []string{"foo"}},
},
structs.CheckServiceNode{
Node: &structs.Node{Node: "node2"},
Service: &structs.NodeService{Tags: []string{"foo", "BAR"}},
},
structs.CheckServiceNode{
Node: &structs.Node{Node: "node3"},
},
structs.CheckServiceNode{
Node: &structs.Node{Node: "node4"},
Service: &structs.NodeService{Tags: []string{"foo", "baz"}},
},
structs.CheckServiceNode{
Node: &structs.Node{Node: "node5"},
Service: &structs.NodeService{Tags: []string{"foo", "zoo"}},
},
structs.CheckServiceNode{
Node: &structs.Node{Node: "node6"},
Service: &structs.NodeService{Tags: []string{"bar"}},
},
}
}
// This always sorts so that it's not annoying to compare after the swap
// operations that the algorithm performs.
stringify := func(nodes structs.CheckServiceNodes) string {
var names []string
for _, node := range nodes {
names = append(names, node.Node.Node)
}
sort.Strings(names)
return strings.Join(names, "|")
}
ret := stringify(tagFilter([]string{}, testNodes()))
if ret != "node1|node2|node3|node4|node5|node6" {
t.Fatalf("bad: %s", ret)
}
ret = stringify(tagFilter([]string{"foo"}, testNodes()))
if ret != "node1|node2|node4|node5" {
t.Fatalf("bad: %s", ret)
}
ret = stringify(tagFilter([]string{"!foo"}, testNodes()))
if ret != "node3|node6" {
t.Fatalf("bad: %s", ret)
}
ret = stringify(tagFilter([]string{"!foo", "bar"}, testNodes()))
if ret != "node6" {
t.Fatalf("bad: %s", ret)
}
ret = stringify(tagFilter([]string{"!foo", "!bar"}, testNodes()))
if ret != "node3" {
t.Fatalf("bad: %s", ret)
}
ret = stringify(tagFilter([]string{"nope"}, testNodes()))
if ret != "" {
t.Fatalf("bad: %s", ret)
}
ret = stringify(tagFilter([]string{"bar"}, testNodes()))
if ret != "node2|node6" {
t.Fatalf("bad: %s", ret)
}
ret = stringify(tagFilter([]string{"BAR"}, testNodes()))
if ret != "node2|node6" {
t.Fatalf("bad: %s", ret)
}
ret = stringify(tagFilter([]string{"bAr"}, testNodes()))
if ret != "node2|node6" {
t.Fatalf("bad: %s", ret)
}
ret = stringify(tagFilter([]string{""}, testNodes()))
if ret != "" {
t.Fatalf("bad: %s", ret)
}
2015-11-11 02:23:37 +00:00
}
2015-11-11 02:30:12 +00:00
func TestPreparedQuery_Wrapper(t *testing.T) {
t.Parallel()
2015-11-11 02:30:12 +00:00
dir1, s1 := testServerWithConfig(t, func(c *Config) {
c.ACLDatacenter = "dc1"
New ACLs (#4791) This PR is almost a complete rewrite of the ACL system within Consul. It brings the features more in line with other HashiCorp products. Obviously there is quite a bit left to do here but most of it is related docs, testing and finishing the last few commands in the CLI. I will update the PR description and check off the todos as I finish them over the next few days/week. Description At a high level this PR is mainly to split ACL tokens from Policies and to split the concepts of Authorization from Identities. A lot of this PR is mostly just to support CRUD operations on ACLTokens and ACLPolicies. These in and of themselves are not particularly interesting. The bigger conceptual changes are in how tokens get resolved, how backwards compatibility is handled and the separation of policy from identity which could lead the way to allowing for alternative identity providers. On the surface and with a new cluster the ACL system will look very similar to that of Nomads. Both have tokens and policies. Both have local tokens. The ACL management APIs for both are very similar. I even ripped off Nomad's ACL bootstrap resetting procedure. There are a few key differences though. Nomad requires token and policy replication where Consul only requires policy replication with token replication being opt-in. In Consul local tokens only work with token replication being enabled though. All policies in Nomad are globally applicable. In Consul all policies are stored and replicated globally but can be scoped to a subset of the datacenters. This allows for more granular access management. Unlike Nomad, Consul has legacy baggage in the form of the original ACL system. The ramifications of this are: A server running the new system must still support other clients using the legacy system. A client running the new system must be able to use the legacy RPCs when the servers in its datacenter are running the legacy system. The primary ACL DC's servers running in legacy mode needs to be a gate that keeps everything else in the entire multi-DC cluster running in legacy mode. So not only does this PR implement the new ACL system but has a legacy mode built in for when the cluster isn't ready for new ACLs. Also detecting that new ACLs can be used is automatic and requires no configuration on the part of administrators. This process is detailed more in the "Transitioning from Legacy to New ACL Mode" section below.
2018-10-19 16:04:07 +00:00
c.ACLsEnabled = true
2015-11-11 02:30:12 +00:00
c.ACLMasterToken = "root"
c.ACLDefaultPolicy = "deny"
})
defer os.RemoveAll(dir1)
defer s1.Shutdown()
dir2, s2 := testServerWithConfig(t, func(c *Config) {
c.Datacenter = "dc2"
c.ACLDatacenter = "dc1"
New ACLs (#4791) This PR is almost a complete rewrite of the ACL system within Consul. It brings the features more in line with other HashiCorp products. Obviously there is quite a bit left to do here but most of it is related docs, testing and finishing the last few commands in the CLI. I will update the PR description and check off the todos as I finish them over the next few days/week. Description At a high level this PR is mainly to split ACL tokens from Policies and to split the concepts of Authorization from Identities. A lot of this PR is mostly just to support CRUD operations on ACLTokens and ACLPolicies. These in and of themselves are not particularly interesting. The bigger conceptual changes are in how tokens get resolved, how backwards compatibility is handled and the separation of policy from identity which could lead the way to allowing for alternative identity providers. On the surface and with a new cluster the ACL system will look very similar to that of Nomads. Both have tokens and policies. Both have local tokens. The ACL management APIs for both are very similar. I even ripped off Nomad's ACL bootstrap resetting procedure. There are a few key differences though. Nomad requires token and policy replication where Consul only requires policy replication with token replication being opt-in. In Consul local tokens only work with token replication being enabled though. All policies in Nomad are globally applicable. In Consul all policies are stored and replicated globally but can be scoped to a subset of the datacenters. This allows for more granular access management. Unlike Nomad, Consul has legacy baggage in the form of the original ACL system. The ramifications of this are: A server running the new system must still support other clients using the legacy system. A client running the new system must be able to use the legacy RPCs when the servers in its datacenter are running the legacy system. The primary ACL DC's servers running in legacy mode needs to be a gate that keeps everything else in the entire multi-DC cluster running in legacy mode. So not only does this PR implement the new ACL system but has a legacy mode built in for when the cluster isn't ready for new ACLs. Also detecting that new ACLs can be used is automatic and requires no configuration on the part of administrators. This process is detailed more in the "Transitioning from Legacy to New ACL Mode" section below.
2018-10-19 16:04:07 +00:00
c.ACLsEnabled = true
2015-11-11 02:30:12 +00:00
c.ACLMasterToken = "root"
c.ACLDefaultPolicy = "deny"
})
defer os.RemoveAll(dir2)
defer s2.Shutdown()
s2.tokens.UpdateReplicationToken("root", tokenStore.TokenSourceConfig)
testrpc.WaitForLeader(t, s1.RPC, "dc1", testrpc.WithToken("root"))
testrpc.WaitForLeader(t, s2.RPC, "dc2", testrpc.WithToken("root"))
2015-11-11 02:30:12 +00:00
// Try to WAN join.
2017-05-05 10:29:49 +00:00
joinWAN(t, s2, s1)
2015-11-11 02:30:12 +00:00
// Try all the operations on a real server via the wrapper.
wrapper := &queryServerWrapper{s1}
wrapper.GetLogger().Debug("Test")
2015-11-11 02:30:12 +00:00
ret, err := wrapper.GetOtherDatacentersByDistance()
wrapper.GetLogger().Info("Returned value", "value", ret)
2015-11-11 02:30:12 +00:00
if err != nil {
t.Fatalf("err: %v", err)
}
if len(ret) != 1 || ret[0] != "dc2" {
t.Fatalf("bad: %v", ret)
}
// Since we have no idea when the joinWAN operation completes
// we keep on querying until the the join operation completes.
retry.Run(t, func(r *retry.R) {
r.Check(s1.forwardDC("Status.Ping", "dc2", &struct{}{}, &struct{}{}))
})
2015-11-11 02:30:12 +00:00
}
type mockQueryServer struct {
Datacenters []string
DatacentersError error
QueryLog []string
QueryFn func(dc string, args interface{}, reply interface{}) error
Logger hclog.Logger
LogBuffer *bytes.Buffer
}
func (m *mockQueryServer) JoinQueryLog() string {
return strings.Join(m.QueryLog, "|")
}
func (m *mockQueryServer) GetLogger() hclog.Logger {
if m.Logger == nil {
m.LogBuffer = new(bytes.Buffer)
m.Logger = hclog.New(&hclog.LoggerOptions{
Name: "mock_query",
Output: m.LogBuffer,
Level: hclog.Debug,
})
}
return m.Logger
}
func (m *mockQueryServer) GetOtherDatacentersByDistance() ([]string, error) {
return m.Datacenters, m.DatacentersError
}
func (m *mockQueryServer) ForwardDC(method, dc string, args interface{}, reply interface{}) error {
m.QueryLog = append(m.QueryLog, fmt.Sprintf("%s:%s", dc, method))
if ret, ok := reply.(*structs.PreparedQueryExecuteResponse); ok {
ret.Datacenter = dc
}
if m.QueryFn != nil {
return m.QueryFn(dc, args, reply)
}
return nil
}
func TestPreparedQuery_queryFailover(t *testing.T) {
t.Parallel()
query := &structs.PreparedQuery{
Name: "test",
Service: structs.ServiceQuery{
Failover: structs.QueryDatacenterOptions{
NearestN: 0,
Datacenters: []string{""},
},
},
}
nodes := func() structs.CheckServiceNodes {
return structs.CheckServiceNodes{
structs.CheckServiceNode{
Node: &structs.Node{Node: "node1"},
},
structs.CheckServiceNode{
Node: &structs.Node{Node: "node2"},
},
structs.CheckServiceNode{
Node: &structs.Node{Node: "node3"},
},
}
}
// Datacenters are available but the query doesn't use them.
{
mock := &mockQueryServer{
Datacenters: []string{"dc1", "dc2", "dc3", "xxx", "dc4"},
}
var reply structs.PreparedQueryExecuteResponse
if err := queryFailover(mock, query, &structs.PreparedQueryExecuteRequest{}, &reply); err != nil {
t.Fatalf("err: %v", err)
}
if len(reply.Nodes) != 0 || reply.Datacenter != "" || reply.Failovers != 0 {
t.Fatalf("bad: %v", reply)
}
}
// Make it fail to get datacenters.
{
mock := &mockQueryServer{
Datacenters: []string{"dc1", "dc2", "dc3", "xxx", "dc4"},
DatacentersError: fmt.Errorf("XXX"),
}
var reply structs.PreparedQueryExecuteResponse
err := queryFailover(mock, query, &structs.PreparedQueryExecuteRequest{}, &reply)
if err == nil || !strings.Contains(err.Error(), "XXX") {
t.Fatalf("bad: %v", err)
}
if len(reply.Nodes) != 0 || reply.Datacenter != "" || reply.Failovers != 0 {
t.Fatalf("bad: %v", reply)
}
}
// The query wants to use other datacenters but none are available.
query.Service.Failover.NearestN = 3
{
mock := &mockQueryServer{
Datacenters: []string{},
}
var reply structs.PreparedQueryExecuteResponse
if err := queryFailover(mock, query, &structs.PreparedQueryExecuteRequest{}, &reply); err != nil {
t.Fatalf("err: %v", err)
}
if len(reply.Nodes) != 0 || reply.Datacenter != "" || reply.Failovers != 0 {
t.Fatalf("bad: %v", reply)
}
}
// Try the first three nearest datacenters, first one has the data.
query.Service.Failover.NearestN = 3
{
mock := &mockQueryServer{
Datacenters: []string{"dc1", "dc2", "dc3", "xxx", "dc4"},
QueryFn: func(dc string, args interface{}, reply interface{}) error {
ret := reply.(*structs.PreparedQueryExecuteResponse)
if dc == "dc1" {
ret.Nodes = nodes()
}
return nil
},
}
var reply structs.PreparedQueryExecuteResponse
if err := queryFailover(mock, query, &structs.PreparedQueryExecuteRequest{}, &reply); err != nil {
t.Fatalf("err: %v", err)
}
if len(reply.Nodes) != 3 ||
reply.Datacenter != "dc1" || reply.Failovers != 1 ||
!reflect.DeepEqual(reply.Nodes, nodes()) {
t.Fatalf("bad: %v", reply)
}
if queries := mock.JoinQueryLog(); queries != "dc1:PreparedQuery.ExecuteRemote" {
t.Fatalf("bad: %s", queries)
}
}
// Try the first three nearest datacenters, last one has the data.
query.Service.Failover.NearestN = 3
{
mock := &mockQueryServer{
Datacenters: []string{"dc1", "dc2", "dc3", "xxx", "dc4"},
QueryFn: func(dc string, args interface{}, reply interface{}) error {
ret := reply.(*structs.PreparedQueryExecuteResponse)
if dc == "dc3" {
ret.Nodes = nodes()
}
return nil
},
}
var reply structs.PreparedQueryExecuteResponse
if err := queryFailover(mock, query, &structs.PreparedQueryExecuteRequest{}, &reply); err != nil {
t.Fatalf("err: %v", err)
}
if len(reply.Nodes) != 3 ||
reply.Datacenter != "dc3" || reply.Failovers != 3 ||
!reflect.DeepEqual(reply.Nodes, nodes()) {
t.Fatalf("bad: %v", reply)
}
if queries := mock.JoinQueryLog(); queries != "dc1:PreparedQuery.ExecuteRemote|dc2:PreparedQuery.ExecuteRemote|dc3:PreparedQuery.ExecuteRemote" {
t.Fatalf("bad: %s", queries)
}
}
// Try the first four nearest datacenters, nobody has the data.
query.Service.Failover.NearestN = 4
{
mock := &mockQueryServer{
Datacenters: []string{"dc1", "dc2", "dc3", "xxx", "dc4"},
}
var reply structs.PreparedQueryExecuteResponse
if err := queryFailover(mock, query, &structs.PreparedQueryExecuteRequest{}, &reply); err != nil {
t.Fatalf("err: %v", err)
}
if len(reply.Nodes) != 0 ||
reply.Datacenter != "xxx" || reply.Failovers != 4 {
t.Fatalf("bad: %v", reply)
}
if queries := mock.JoinQueryLog(); queries != "dc1:PreparedQuery.ExecuteRemote|dc2:PreparedQuery.ExecuteRemote|dc3:PreparedQuery.ExecuteRemote|xxx:PreparedQuery.ExecuteRemote" {
t.Fatalf("bad: %s", queries)
}
}
// Try the first two nearest datacenters, plus a user-specified one that
// has the data.
query.Service.Failover.NearestN = 2
query.Service.Failover.Datacenters = []string{"dc4"}
{
mock := &mockQueryServer{
Datacenters: []string{"dc1", "dc2", "dc3", "xxx", "dc4"},
QueryFn: func(dc string, args interface{}, reply interface{}) error {
ret := reply.(*structs.PreparedQueryExecuteResponse)
if dc == "dc4" {
ret.Nodes = nodes()
}
return nil
},
}
var reply structs.PreparedQueryExecuteResponse
if err := queryFailover(mock, query, &structs.PreparedQueryExecuteRequest{}, &reply); err != nil {
t.Fatalf("err: %v", err)
}
if len(reply.Nodes) != 3 ||
reply.Datacenter != "dc4" || reply.Failovers != 3 ||
!reflect.DeepEqual(reply.Nodes, nodes()) {
t.Fatalf("bad: %v", reply)
}
if queries := mock.JoinQueryLog(); queries != "dc1:PreparedQuery.ExecuteRemote|dc2:PreparedQuery.ExecuteRemote|dc4:PreparedQuery.ExecuteRemote" {
t.Fatalf("bad: %s", queries)
}
}
// Add in a hard-coded value that overlaps with the nearest list.
query.Service.Failover.NearestN = 2
query.Service.Failover.Datacenters = []string{"dc4", "dc1"}
{
mock := &mockQueryServer{
Datacenters: []string{"dc1", "dc2", "dc3", "xxx", "dc4"},
QueryFn: func(dc string, args interface{}, reply interface{}) error {
ret := reply.(*structs.PreparedQueryExecuteResponse)
if dc == "dc4" {
ret.Nodes = nodes()
}
return nil
},
}
var reply structs.PreparedQueryExecuteResponse
if err := queryFailover(mock, query, &structs.PreparedQueryExecuteRequest{}, &reply); err != nil {
t.Fatalf("err: %v", err)
}
if len(reply.Nodes) != 3 ||
reply.Datacenter != "dc4" || reply.Failovers != 3 ||
!reflect.DeepEqual(reply.Nodes, nodes()) {
t.Fatalf("bad: %v", reply)
}
if queries := mock.JoinQueryLog(); queries != "dc1:PreparedQuery.ExecuteRemote|dc2:PreparedQuery.ExecuteRemote|dc4:PreparedQuery.ExecuteRemote" {
t.Fatalf("bad: %s", queries)
}
}
// Now add a bogus user-defined one to the mix.
query.Service.Failover.NearestN = 2
query.Service.Failover.Datacenters = []string{"nope", "dc4", "dc1"}
{
mock := &mockQueryServer{
Datacenters: []string{"dc1", "dc2", "dc3", "xxx", "dc4"},
QueryFn: func(dc string, args interface{}, reply interface{}) error {
ret := reply.(*structs.PreparedQueryExecuteResponse)
if dc == "dc4" {
ret.Nodes = nodes()
}
return nil
},
}
var reply structs.PreparedQueryExecuteResponse
if err := queryFailover(mock, query, &structs.PreparedQueryExecuteRequest{}, &reply); err != nil {
t.Fatalf("err: %v", err)
}
if len(reply.Nodes) != 3 ||
reply.Datacenter != "dc4" || reply.Failovers != 3 ||
!reflect.DeepEqual(reply.Nodes, nodes()) {
t.Fatalf("bad: %v", reply)
}
if queries := mock.JoinQueryLog(); queries != "dc1:PreparedQuery.ExecuteRemote|dc2:PreparedQuery.ExecuteRemote|dc4:PreparedQuery.ExecuteRemote" {
t.Fatalf("bad: %s", queries)
}
if !strings.Contains(mock.LogBuffer.String(), "Skipping unknown datacenter") {
t.Fatalf("bad: %s", mock.LogBuffer.String())
}
}
// Same setup as before but dc1 is going to return an error and should
// get skipped over, still yielding data from dc4 which comes later.
query.Service.Failover.NearestN = 2
query.Service.Failover.Datacenters = []string{"dc4", "dc1"}
{
mock := &mockQueryServer{
Datacenters: []string{"dc1", "dc2", "dc3", "xxx", "dc4"},
QueryFn: func(dc string, args interface{}, reply interface{}) error {
ret := reply.(*structs.PreparedQueryExecuteResponse)
if dc == "dc1" {
return fmt.Errorf("XXX")
} else if dc == "dc4" {
ret.Nodes = nodes()
}
return nil
},
}
var reply structs.PreparedQueryExecuteResponse
if err := queryFailover(mock, query, &structs.PreparedQueryExecuteRequest{}, &reply); err != nil {
t.Fatalf("err: %v", err)
}
if len(reply.Nodes) != 3 ||
reply.Datacenter != "dc4" || reply.Failovers != 3 ||
!reflect.DeepEqual(reply.Nodes, nodes()) {
t.Fatalf("bad: %v", reply)
}
if queries := mock.JoinQueryLog(); queries != "dc1:PreparedQuery.ExecuteRemote|dc2:PreparedQuery.ExecuteRemote|dc4:PreparedQuery.ExecuteRemote" {
t.Fatalf("bad: %s", queries)
}
if !strings.Contains(mock.LogBuffer.String(), "Failed querying") {
t.Fatalf("bad: %s", mock.LogBuffer.String())
}
}
// Just use a hard-coded list and now xxx has the data.
query.Service.Failover.NearestN = 0
query.Service.Failover.Datacenters = []string{"dc3", "xxx"}
{
mock := &mockQueryServer{
Datacenters: []string{"dc1", "dc2", "dc3", "xxx", "dc4"},
QueryFn: func(dc string, args interface{}, reply interface{}) error {
ret := reply.(*structs.PreparedQueryExecuteResponse)
if dc == "xxx" {
ret.Nodes = nodes()
}
return nil
},
}
var reply structs.PreparedQueryExecuteResponse
if err := queryFailover(mock, query, &structs.PreparedQueryExecuteRequest{}, &reply); err != nil {
t.Fatalf("err: %v", err)
}
if len(reply.Nodes) != 3 ||
reply.Datacenter != "xxx" || reply.Failovers != 2 ||
!reflect.DeepEqual(reply.Nodes, nodes()) {
t.Fatalf("bad: %v", reply)
}
if queries := mock.JoinQueryLog(); queries != "dc3:PreparedQuery.ExecuteRemote|xxx:PreparedQuery.ExecuteRemote" {
t.Fatalf("bad: %s", queries)
}
}
// Make sure the limit and query options are plumbed through.
query.Service.Failover.NearestN = 0
query.Service.Failover.Datacenters = []string{"xxx"}
{
mock := &mockQueryServer{
Datacenters: []string{"dc1", "dc2", "dc3", "xxx", "dc4"},
QueryFn: func(dc string, args interface{}, reply interface{}) error {
inp := args.(*structs.PreparedQueryExecuteRemoteRequest)
ret := reply.(*structs.PreparedQueryExecuteResponse)
if dc == "xxx" {
if inp.Limit != 5 {
t.Fatalf("bad: %d", inp.Limit)
}
if inp.RequireConsistent != true {
t.Fatalf("bad: %v", inp.RequireConsistent)
}
ret.Nodes = nodes()
}
return nil
},
}
var reply structs.PreparedQueryExecuteResponse
if err := queryFailover(mock, query, &structs.PreparedQueryExecuteRequest{
Limit: 5,
QueryOptions: structs.QueryOptions{RequireConsistent: true},
}, &reply); err != nil {
t.Fatalf("err: %v", err)
}
if len(reply.Nodes) != 3 ||
reply.Datacenter != "xxx" || reply.Failovers != 1 ||
!reflect.DeepEqual(reply.Nodes, nodes()) {
t.Fatalf("bad: %v", reply)
}
if queries := mock.JoinQueryLog(); queries != "xxx:PreparedQuery.ExecuteRemote" {
t.Fatalf("bad: %s", queries)
}
}
}