consul/acl/policy.go

851 lines
24 KiB
Go
Raw Normal View History

2014-08-06 22:08:17 +00:00
package acl
import (
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
"bytes"
"encoding/binary"
2014-08-06 22:08:17 +00:00
"fmt"
2014-12-01 03:18:16 +00:00
"github.com/hashicorp/consul/sentinel"
2014-08-06 22:08:17 +00:00
"github.com/hashicorp/hcl"
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
"github.com/hashicorp/hcl/hcl/ast"
hclprinter "github.com/hashicorp/hcl/hcl/printer"
"golang.org/x/crypto/blake2b"
)
type SyntaxVersion int
const (
SyntaxCurrent SyntaxVersion = iota
SyntaxLegacy
2014-08-06 22:08:17 +00:00
)
const (
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
PolicyDeny = "deny"
PolicyRead = "read"
PolicyWrite = "write"
PolicyList = "list"
2014-08-06 22:08:17 +00:00
)
// Policy is used to represent the policy specified by
// an ACL configuration.
type Policy struct {
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
ID string `hcl:"id"`
Revision uint64 `hcl:"revision"`
ACL string `hcl:"acl,expand"`
Agents []*AgentPolicy `hcl:"agent,expand"`
AgentPrefixes []*AgentPolicy `hcl:"agent_prefix,expand"`
Keys []*KeyPolicy `hcl:"key,expand"`
KeyPrefixes []*KeyPolicy `hcl:"key_prefix,expand"`
Nodes []*NodePolicy `hcl:"node,expand"`
NodePrefixes []*NodePolicy `hcl:"node_prefix,expand"`
Services []*ServicePolicy `hcl:"service,expand"`
ServicePrefixes []*ServicePolicy `hcl:"service_prefix,expand"`
Sessions []*SessionPolicy `hcl:"session,expand"`
SessionPrefixes []*SessionPolicy `hcl:"session_prefix,expand"`
Events []*EventPolicy `hcl:"event,expand"`
EventPrefixes []*EventPolicy `hcl:"event_prefix,expand"`
PreparedQueries []*PreparedQueryPolicy `hcl:"query,expand"`
PreparedQueryPrefixes []*PreparedQueryPolicy `hcl:"query_prefix,expand"`
Keyring string `hcl:"keyring"`
Operator string `hcl:"operator"`
2014-08-06 22:08:17 +00:00
}
// Sentinel defines a snippet of Sentinel code that can be attached to a policy.
type Sentinel struct {
Code string
EnforcementLevel string
}
// AgentPolicy represents a policy for working with agent endpoints on nodes
// with specific name prefixes.
type AgentPolicy struct {
Node string `hcl:",key"`
Policy string
}
func (a *AgentPolicy) GoString() string {
return fmt.Sprintf("%#v", *a)
}
2014-08-06 22:08:17 +00:00
// KeyPolicy represents a policy for a key
type KeyPolicy struct {
Prefix string `hcl:",key"`
Policy string
Sentinel Sentinel
2014-08-08 22:57:28 +00:00
}
func (k *KeyPolicy) GoString() string {
return fmt.Sprintf("%#v", *k)
2014-08-06 22:08:17 +00:00
}
// NodePolicy represents a policy for a node
type NodePolicy struct {
Name string `hcl:",key"`
Policy string
Sentinel Sentinel
}
func (n *NodePolicy) GoString() string {
return fmt.Sprintf("%#v", *n)
}
2014-12-01 03:18:16 +00:00
// ServicePolicy represents a policy for a service
type ServicePolicy struct {
Name string `hcl:",key"`
Policy string
Sentinel Sentinel
// Intentions is the policy for intentions where this service is the
// destination. This may be empty, in which case the Policy determines
// the intentions policy.
Intentions string
2014-12-01 03:18:16 +00:00
}
func (s *ServicePolicy) GoString() string {
return fmt.Sprintf("%#v", *s)
2014-12-01 03:18:16 +00:00
}
// SessionPolicy represents a policy for making sessions tied to specific node
// name prefixes.
type SessionPolicy struct {
Node string `hcl:",key"`
Policy string
}
func (s *SessionPolicy) GoString() string {
return fmt.Sprintf("%#v", *s)
}
2015-06-18 01:56:29 +00:00
// EventPolicy represents a user event policy.
type EventPolicy struct {
Event string `hcl:",key"`
Policy string
}
func (e *EventPolicy) GoString() string {
return fmt.Sprintf("%#v", *e)
}
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
// PreparedQueryPolicy represents a prepared query policy.
type PreparedQueryPolicy struct {
Prefix string `hcl:",key"`
Policy string
}
func (p *PreparedQueryPolicy) GoString() string {
return fmt.Sprintf("%#v", *p)
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
}
// isPolicyValid makes sure the given string matches one of the valid policies.
func isPolicyValid(policy string) bool {
switch policy {
case PolicyDeny:
return true
case PolicyRead:
return true
case PolicyWrite:
return true
default:
return false
}
}
// isSentinelValid makes sure the given sentinel block is valid, and will skip
// out if the evaluator is nil.
func isSentinelValid(sentinel sentinel.Evaluator, basicPolicy string, sp Sentinel) error {
// Sentinel not enabled at all, or for this policy.
if sentinel == nil {
return nil
}
if sp.Code == "" {
return nil
}
// We only allow sentinel code on write policies at this time.
if basicPolicy != PolicyWrite {
return fmt.Errorf("code is only allowed for write policies")
}
// Validate the sentinel parts.
switch sp.EnforcementLevel {
case "", "soft-mandatory", "hard-mandatory":
// OK
default:
return fmt.Errorf("unsupported enforcement level %q", sp.EnforcementLevel)
}
return sentinel.Compile(sp.Code)
}
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
func parseCurrent(rules string, sentinel sentinel.Evaluator) (*Policy, error) {
2014-08-06 22:08:17 +00:00
p := &Policy{}
2014-08-08 21:36:09 +00:00
2014-08-06 22:08:17 +00:00
if err := hcl.Decode(p, rules); err != nil {
return nil, fmt.Errorf("Failed to parse ACL rules: %v", err)
}
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
// Validate the acl policy
if p.ACL != "" && !isPolicyValid(p.ACL) {
return nil, fmt.Errorf("Invalid acl policy: %#v", p.ACL)
}
// Validate the agent policy
for _, ap := range p.Agents {
if !isPolicyValid(ap.Policy) {
return nil, fmt.Errorf("Invalid agent policy: %#v", ap)
}
}
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
for _, ap := range p.AgentPrefixes {
if !isPolicyValid(ap.Policy) {
return nil, fmt.Errorf("Invalid agent_prefix policy: %#v", ap)
}
}
2014-08-06 22:08:17 +00:00
// Validate the key policy
for _, kp := range p.Keys {
if kp.Policy != PolicyList && !isPolicyValid(kp.Policy) {
2014-08-06 22:08:17 +00:00
return nil, fmt.Errorf("Invalid key policy: %#v", kp)
}
if err := isSentinelValid(sentinel, kp.Policy, kp.Sentinel); err != nil {
return nil, fmt.Errorf("Invalid key Sentinel policy: %#v, got error:%v", kp, err)
}
2014-08-06 22:08:17 +00:00
}
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
for _, kp := range p.KeyPrefixes {
if kp.Policy != PolicyList && !isPolicyValid(kp.Policy) {
return nil, fmt.Errorf("Invalid key_prefix policy: %#v", kp)
}
if err := isSentinelValid(sentinel, kp.Policy, kp.Sentinel); err != nil {
return nil, fmt.Errorf("Invalid key_prefix Sentinel policy: %#v, got error:%v", kp, err)
}
}
2014-12-01 03:18:16 +00:00
// Validate the node policies
for _, np := range p.Nodes {
if !isPolicyValid(np.Policy) {
return nil, fmt.Errorf("Invalid node policy: %#v", np)
}
if err := isSentinelValid(sentinel, np.Policy, np.Sentinel); err != nil {
return nil, fmt.Errorf("Invalid node Sentinel policy: %#v, got error:%v", np, err)
}
}
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
for _, np := range p.NodePrefixes {
if !isPolicyValid(np.Policy) {
return nil, fmt.Errorf("Invalid node_prefix policy: %#v", np)
}
if err := isSentinelValid(sentinel, np.Policy, np.Sentinel); err != nil {
return nil, fmt.Errorf("Invalid node_prefix Sentinel policy: %#v, got error:%v", np, err)
}
}
// Validate the service policies
2014-12-01 03:18:16 +00:00
for _, sp := range p.Services {
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 !isPolicyValid(sp.Policy) {
2014-12-01 03:18:16 +00:00
return nil, fmt.Errorf("Invalid service policy: %#v", sp)
}
if sp.Intentions != "" && !isPolicyValid(sp.Intentions) {
return nil, fmt.Errorf("Invalid service intentions policy: %#v", sp)
}
if err := isSentinelValid(sentinel, sp.Policy, sp.Sentinel); err != nil {
return nil, fmt.Errorf("Invalid service Sentinel policy: %#v, got error:%v", sp, err)
}
2014-12-01 03:18:16 +00:00
}
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
for _, sp := range p.ServicePrefixes {
if !isPolicyValid(sp.Policy) {
return nil, fmt.Errorf("Invalid service_prefix policy: %#v", sp)
}
if sp.Intentions != "" && !isPolicyValid(sp.Intentions) {
return nil, fmt.Errorf("Invalid service_prefix intentions policy: %#v", sp)
}
if err := isSentinelValid(sentinel, sp.Policy, sp.Sentinel); err != nil {
return nil, fmt.Errorf("Invalid service_prefix Sentinel policy: %#v, got error:%v", sp, err)
}
}
2014-12-01 03:18:16 +00:00
// Validate the session policies
for _, sp := range p.Sessions {
if !isPolicyValid(sp.Policy) {
return nil, fmt.Errorf("Invalid session policy: %#v", sp)
}
}
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
for _, sp := range p.SessionPrefixes {
if !isPolicyValid(sp.Policy) {
return nil, fmt.Errorf("Invalid session_prefix policy: %#v", sp)
}
}
2015-06-18 01:56:29 +00:00
// Validate the user event policies
for _, ep := range p.Events {
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 !isPolicyValid(ep.Policy) {
2015-06-18 01:56:29 +00:00
return nil, fmt.Errorf("Invalid event policy: %#v", ep)
}
}
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
for _, ep := range p.EventPrefixes {
if !isPolicyValid(ep.Policy) {
return nil, fmt.Errorf("Invalid event_prefix policy: %#v", ep)
}
}
2015-06-18 01:56:29 +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
// Validate the prepared query policies
for _, pq := range p.PreparedQueries {
if !isPolicyValid(pq.Policy) {
return nil, fmt.Errorf("Invalid query policy: %#v", pq)
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
}
}
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
for _, pq := range p.PreparedQueryPrefixes {
if !isPolicyValid(pq.Policy) {
return nil, fmt.Errorf("Invalid query_prefix policy: %#v", pq)
}
}
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
// Validate the keyring policy - this one is allowed to be empty
if p.Keyring != "" && !isPolicyValid(p.Keyring) {
2015-07-07 16:45:38 +00:00
return nil, fmt.Errorf("Invalid keyring policy: %#v", p.Keyring)
2015-07-07 00:28:09 +00:00
}
// Validate the operator policy - this one is allowed to be empty
if p.Operator != "" && !isPolicyValid(p.Operator) {
return nil, fmt.Errorf("Invalid operator policy: %#v", p.Operator)
}
2014-08-06 22:08:17 +00:00
return p, nil
}
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
func parseLegacy(rules string, sentinel sentinel.Evaluator) (*Policy, error) {
p := &Policy{}
type LegacyPolicy struct {
Agents []*AgentPolicy `hcl:"agent,expand"`
Keys []*KeyPolicy `hcl:"key,expand"`
Nodes []*NodePolicy `hcl:"node,expand"`
Services []*ServicePolicy `hcl:"service,expand"`
Sessions []*SessionPolicy `hcl:"session,expand"`
Events []*EventPolicy `hcl:"event,expand"`
PreparedQueries []*PreparedQueryPolicy `hcl:"query,expand"`
Keyring string `hcl:"keyring"`
Operator string `hcl:"operator"`
}
lp := &LegacyPolicy{}
if err := hcl.Decode(lp, rules); err != nil {
return nil, fmt.Errorf("Failed to parse ACL rules: %v", err)
}
// Validate the agent policy
for _, ap := range lp.Agents {
if !isPolicyValid(ap.Policy) {
return nil, fmt.Errorf("Invalid agent policy: %#v", ap)
}
p.AgentPrefixes = append(p.AgentPrefixes, ap)
}
// Validate the key policy
for _, kp := range lp.Keys {
if kp.Policy != PolicyList && !isPolicyValid(kp.Policy) {
return nil, fmt.Errorf("Invalid key policy: %#v", kp)
}
if err := isSentinelValid(sentinel, kp.Policy, kp.Sentinel); err != nil {
return nil, fmt.Errorf("Invalid key Sentinel policy: %#v, got error:%v", kp, err)
}
p.KeyPrefixes = append(p.KeyPrefixes, kp)
}
// Validate the node policies
for _, np := range lp.Nodes {
if !isPolicyValid(np.Policy) {
return nil, fmt.Errorf("Invalid node policy: %#v", np)
}
if err := isSentinelValid(sentinel, np.Policy, np.Sentinel); err != nil {
return nil, fmt.Errorf("Invalid node Sentinel policy: %#v, got error:%v", np, err)
}
p.NodePrefixes = append(p.NodePrefixes, np)
}
// Validate the service policies
for _, sp := range lp.Services {
if !isPolicyValid(sp.Policy) {
return nil, fmt.Errorf("Invalid service policy: %#v", sp)
}
if sp.Intentions != "" && !isPolicyValid(sp.Intentions) {
return nil, fmt.Errorf("Invalid service intentions policy: %#v", sp)
}
if err := isSentinelValid(sentinel, sp.Policy, sp.Sentinel); err != nil {
return nil, fmt.Errorf("Invalid service Sentinel policy: %#v, got error:%v", sp, err)
}
p.ServicePrefixes = append(p.ServicePrefixes, sp)
}
// Validate the session policies
for _, sp := range lp.Sessions {
if !isPolicyValid(sp.Policy) {
return nil, fmt.Errorf("Invalid session policy: %#v", sp)
}
p.SessionPrefixes = append(p.SessionPrefixes, sp)
}
// Validate the user event policies
for _, ep := range lp.Events {
if !isPolicyValid(ep.Policy) {
return nil, fmt.Errorf("Invalid event policy: %#v", ep)
}
p.EventPrefixes = append(p.EventPrefixes, ep)
}
// Validate the prepared query policies
for _, pq := range lp.PreparedQueries {
if !isPolicyValid(pq.Policy) {
return nil, fmt.Errorf("Invalid query policy: %#v", pq)
}
p.PreparedQueryPrefixes = append(p.PreparedQueryPrefixes, pq)
}
// Validate the keyring policy - this one is allowed to be empty
if lp.Keyring != "" && !isPolicyValid(lp.Keyring) {
return nil, fmt.Errorf("Invalid keyring policy: %#v", lp.Keyring)
} else {
p.Keyring = lp.Keyring
}
// Validate the operator policy - this one is allowed to be empty
if lp.Operator != "" && !isPolicyValid(lp.Operator) {
return nil, fmt.Errorf("Invalid operator policy: %#v", lp.Operator)
} else {
p.Operator = lp.Operator
}
return p, nil
}
// NewPolicyFromSource is used to parse the specified ACL rules into an
// intermediary set of policies, before being compiled into
// the ACL
func NewPolicyFromSource(id string, revision uint64, rules string, syntax SyntaxVersion, sentinel sentinel.Evaluator) (*Policy, error) {
if rules == "" {
// Hot path for empty source
return &Policy{ID: id, Revision: revision}, nil
}
var policy *Policy
var err error
switch syntax {
case SyntaxLegacy:
policy, err = parseLegacy(rules, sentinel)
case SyntaxCurrent:
policy, err = parseCurrent(rules, sentinel)
default:
return nil, fmt.Errorf("Invalid rules version: %d", syntax)
}
if err == nil {
policy.ID = id
policy.Revision = revision
}
return policy, err
}
func (policy *Policy) ConvertToLegacy() *Policy {
converted := &Policy{
ID: policy.ID,
Revision: policy.Revision,
ACL: policy.ACL,
Keyring: policy.Keyring,
Operator: policy.Operator,
}
converted.Agents = append(converted.Agents, policy.Agents...)
converted.Agents = append(converted.Agents, policy.AgentPrefixes...)
converted.Keys = append(converted.Keys, policy.Keys...)
converted.Keys = append(converted.Keys, policy.KeyPrefixes...)
converted.Nodes = append(converted.Nodes, policy.Nodes...)
converted.Nodes = append(converted.Nodes, policy.NodePrefixes...)
converted.Services = append(converted.Services, policy.Services...)
converted.Services = append(converted.Services, policy.ServicePrefixes...)
converted.Sessions = append(converted.Sessions, policy.Sessions...)
converted.Sessions = append(converted.Sessions, policy.SessionPrefixes...)
converted.Events = append(converted.Events, policy.Events...)
converted.Events = append(converted.Events, policy.EventPrefixes...)
converted.PreparedQueries = append(converted.PreparedQueries, policy.PreparedQueries...)
converted.PreparedQueries = append(converted.PreparedQueries, policy.PreparedQueryPrefixes...)
return converted
}
func (policy *Policy) ConvertFromLegacy() *Policy {
return &Policy{
ID: policy.ID,
Revision: policy.Revision,
AgentPrefixes: policy.Agents,
KeyPrefixes: policy.Keys,
NodePrefixes: policy.Nodes,
ServicePrefixes: policy.Services,
SessionPrefixes: policy.Sessions,
EventPrefixes: policy.Events,
PreparedQueryPrefixes: policy.PreparedQueries,
Keyring: policy.Keyring,
Operator: policy.Operator,
}
}
// takesPrecedenceOver returns true when permission a
// should take precedence over permission b
func takesPrecedenceOver(a, b string) bool {
if a == PolicyDeny {
return true
} else if b == PolicyDeny {
return false
}
if a == PolicyWrite {
return true
} else if b == PolicyWrite {
return false
}
if a == PolicyList {
return true
} else if b == PolicyList {
return false
}
if a == PolicyRead {
return true
} else if b == PolicyRead {
return false
}
return false
}
func multiPolicyID(policies []*Policy) []byte {
cacheKeyHash, err := blake2b.New256(nil)
if err != nil {
panic(err)
}
for _, policy := range policies {
cacheKeyHash.Write([]byte(policy.ID))
binary.Write(cacheKeyHash, binary.BigEndian, policy.Revision)
}
return cacheKeyHash.Sum(nil)
}
// MergePolicies merges multiple ACL policies into one policy
// This function will not set either the ID or the Scope fields
// of the resulting policy as its up to the caller to determine
// what the merged value is.
func MergePolicies(policies []*Policy) *Policy {
// maps are used here so that we can lookup each policy by
// the segment that the rule applies to during the policy
// merge. Otherwise we could do a linear search through a slice
// and replace it inline
aclPolicy := ""
agentPolicies := make(map[string]*AgentPolicy)
agentPrefixPolicies := make(map[string]*AgentPolicy)
eventPolicies := make(map[string]*EventPolicy)
eventPrefixPolicies := make(map[string]*EventPolicy)
keyringPolicy := ""
keyPolicies := make(map[string]*KeyPolicy)
keyPrefixPolicies := make(map[string]*KeyPolicy)
nodePolicies := make(map[string]*NodePolicy)
nodePrefixPolicies := make(map[string]*NodePolicy)
operatorPolicy := ""
preparedQueryPolicies := make(map[string]*PreparedQueryPolicy)
preparedQueryPrefixPolicies := make(map[string]*PreparedQueryPolicy)
servicePolicies := make(map[string]*ServicePolicy)
servicePrefixPolicies := make(map[string]*ServicePolicy)
sessionPolicies := make(map[string]*SessionPolicy)
sessionPrefixPolicies := make(map[string]*SessionPolicy)
// Parse all the individual rule sets
for _, policy := range policies {
if takesPrecedenceOver(policy.ACL, aclPolicy) {
aclPolicy = policy.ACL
}
for _, ap := range policy.Agents {
update := true
if permission, found := agentPolicies[ap.Node]; found {
update = takesPrecedenceOver(ap.Policy, permission.Policy)
}
if update {
agentPolicies[ap.Node] = ap
}
}
for _, ap := range policy.AgentPrefixes {
update := true
if permission, found := agentPrefixPolicies[ap.Node]; found {
update = takesPrecedenceOver(ap.Policy, permission.Policy)
}
if update {
agentPrefixPolicies[ap.Node] = ap
}
}
for _, ep := range policy.Events {
update := true
if permission, found := eventPolicies[ep.Event]; found {
update = takesPrecedenceOver(ep.Policy, permission.Policy)
}
if update {
eventPolicies[ep.Event] = ep
}
}
for _, ep := range policy.EventPrefixes {
update := true
if permission, found := eventPrefixPolicies[ep.Event]; found {
update = takesPrecedenceOver(ep.Policy, permission.Policy)
}
if update {
eventPrefixPolicies[ep.Event] = ep
}
}
if takesPrecedenceOver(policy.Keyring, keyringPolicy) {
keyringPolicy = policy.Keyring
}
for _, kp := range policy.Keys {
update := true
if permission, found := keyPolicies[kp.Prefix]; found {
update = takesPrecedenceOver(kp.Policy, permission.Policy)
}
if update {
keyPolicies[kp.Prefix] = kp
}
}
for _, kp := range policy.KeyPrefixes {
update := true
if permission, found := keyPrefixPolicies[kp.Prefix]; found {
update = takesPrecedenceOver(kp.Policy, permission.Policy)
}
if update {
keyPrefixPolicies[kp.Prefix] = kp
}
}
for _, np := range policy.Nodes {
update := true
if permission, found := nodePolicies[np.Name]; found {
update = takesPrecedenceOver(np.Policy, permission.Policy)
}
if update {
nodePolicies[np.Name] = np
}
}
for _, np := range policy.NodePrefixes {
update := true
if permission, found := nodePrefixPolicies[np.Name]; found {
update = takesPrecedenceOver(np.Policy, permission.Policy)
}
if update {
nodePrefixPolicies[np.Name] = np
}
}
if takesPrecedenceOver(policy.Operator, operatorPolicy) {
operatorPolicy = policy.Operator
}
for _, qp := range policy.PreparedQueries {
update := true
if permission, found := preparedQueryPolicies[qp.Prefix]; found {
update = takesPrecedenceOver(qp.Policy, permission.Policy)
}
if update {
preparedQueryPolicies[qp.Prefix] = qp
}
}
for _, qp := range policy.PreparedQueryPrefixes {
update := true
if permission, found := preparedQueryPrefixPolicies[qp.Prefix]; found {
update = takesPrecedenceOver(qp.Policy, permission.Policy)
}
if update {
preparedQueryPrefixPolicies[qp.Prefix] = qp
}
}
for _, sp := range policy.Services {
existing, found := servicePolicies[sp.Name]
if !found {
servicePolicies[sp.Name] = sp
continue
}
if takesPrecedenceOver(sp.Policy, existing.Policy) {
existing.Policy = sp.Policy
existing.Sentinel = sp.Sentinel
}
if takesPrecedenceOver(sp.Intentions, existing.Intentions) {
existing.Intentions = sp.Intentions
}
}
for _, sp := range policy.ServicePrefixes {
existing, found := servicePrefixPolicies[sp.Name]
if !found {
servicePrefixPolicies[sp.Name] = sp
continue
}
if takesPrecedenceOver(sp.Policy, existing.Policy) {
existing.Policy = sp.Policy
existing.Sentinel = sp.Sentinel
}
if takesPrecedenceOver(sp.Intentions, existing.Intentions) {
existing.Intentions = sp.Intentions
}
}
for _, sp := range policy.Sessions {
update := true
if permission, found := sessionPolicies[sp.Node]; found {
update = takesPrecedenceOver(sp.Policy, permission.Policy)
}
if update {
sessionPolicies[sp.Node] = sp
}
}
for _, sp := range policy.SessionPrefixes {
update := true
if permission, found := sessionPrefixPolicies[sp.Node]; found {
update = takesPrecedenceOver(sp.Policy, permission.Policy)
}
if update {
sessionPrefixPolicies[sp.Node] = sp
}
}
}
merged := &Policy{ACL: aclPolicy, Keyring: keyringPolicy, Operator: operatorPolicy}
// All the for loop appends are ugly but Go doesn't have a way to get
// a slice of all values within a map so this is necessary
for _, policy := range agentPolicies {
merged.Agents = append(merged.Agents, policy)
}
for _, policy := range agentPrefixPolicies {
merged.AgentPrefixes = append(merged.AgentPrefixes, policy)
}
for _, policy := range eventPolicies {
merged.Events = append(merged.Events, policy)
}
for _, policy := range eventPrefixPolicies {
merged.EventPrefixes = append(merged.EventPrefixes, policy)
}
for _, policy := range keyPolicies {
merged.Keys = append(merged.Keys, policy)
}
for _, policy := range keyPrefixPolicies {
merged.KeyPrefixes = append(merged.KeyPrefixes, policy)
}
for _, policy := range nodePolicies {
merged.Nodes = append(merged.Nodes, policy)
}
for _, policy := range nodePrefixPolicies {
merged.NodePrefixes = append(merged.NodePrefixes, policy)
}
for _, policy := range preparedQueryPolicies {
merged.PreparedQueries = append(merged.PreparedQueries, policy)
}
for _, policy := range preparedQueryPrefixPolicies {
merged.PreparedQueryPrefixes = append(merged.PreparedQueryPrefixes, policy)
}
for _, policy := range servicePolicies {
merged.Services = append(merged.Services, policy)
}
for _, policy := range servicePrefixPolicies {
merged.ServicePrefixes = append(merged.ServicePrefixes, policy)
}
for _, policy := range sessionPolicies {
merged.Sessions = append(merged.Sessions, policy)
}
for _, policy := range sessionPrefixPolicies {
merged.SessionPrefixes = append(merged.SessionPrefixes, policy)
}
merged.ID = fmt.Sprintf("%x", multiPolicyID(policies))
return merged
}
func TranslateLegacyRules(policyBytes []byte) ([]byte, error) {
parsed, err := hcl.ParseBytes(policyBytes)
if err != nil {
return nil, fmt.Errorf("Failed to parse rules: %v", err)
}
rewritten := ast.Walk(parsed, func(node ast.Node) (ast.Node, bool) {
switch n := node.(type) {
case *ast.ObjectKey:
switch n.Token.Text {
case "agent":
n.Token.Text = "agent_prefix"
case "key":
n.Token.Text = "key_prefix"
case "node":
n.Token.Text = "node_prefix"
case "query":
n.Token.Text = "query_prefix"
case "service":
n.Token.Text = "service_prefix"
case "session":
n.Token.Text = "session_prefix"
case "event":
n.Token.Text = "event_prefix"
}
}
return node, true
})
buffer := new(bytes.Buffer)
if err := hclprinter.Fprint(buffer, rewritten); err != nil {
return nil, fmt.Errorf("Failed to output new rules: %v", err)
}
return buffer.Bytes(), nil
}