consul/acl/acl.go

550 lines
12 KiB
Go
Raw Normal View History

2014-08-06 22:08:17 +00:00
package acl
import (
"github.com/armon/go-radix"
)
var (
// allowAll is a singleton policy which allows all
// non-management actions
2014-08-06 22:08:17 +00:00
allowAll ACL
// denyAll is a singleton policy which denies all actions
denyAll ACL
// manageAll is a singleton policy which allows all
// actions, including management
manageAll ACL
2014-08-06 22:08:17 +00:00
)
func init() {
// Setup the singletons
allowAll = &StaticACL{
allowManage: false,
defaultAllow: true,
}
denyAll = &StaticACL{
allowManage: false,
defaultAllow: false,
}
manageAll = &StaticACL{
allowManage: true,
defaultAllow: true,
}
2014-08-06 22:08:17 +00:00
}
// ACL is the interface for policy enforcement.
type ACL interface {
// ACLList checks for permission to list all the ACLs
ACLList() bool
// ACLModify checks for permission to manipulate ACLs
ACLModify() bool
// EventRead determines if a specific event can be queried.
EventRead(string) bool
// EventWrite determines if a specific event may be fired.
EventWrite(string) bool
// KeyRead checks for permission to read a given key
2014-08-06 22:08:17 +00:00
KeyRead(string) bool
// KeyWrite checks for permission to write a given key
2014-08-06 22:08:17 +00:00
KeyWrite(string) bool
// KeyWritePrefix checks for permission to write to an
// entire key prefix. This means there must be no sub-policies
// that deny a write.
KeyWritePrefix(string) bool
2015-07-07 00:28:09 +00:00
// KeyringRead determines if the encryption keyring used in
// the gossip layer can be read.
KeyringRead() bool
// KeyringWrite determines if the keyring can be manipulated
KeyringWrite() bool
// NodeRead checks for permission to read (discover) a given node.
NodeRead(string) bool
// NodeWrite checks for permission to create or update (register) a
// given node.
NodeWrite(string) bool
// OperatorRead determines if the read-only Consul operator functions
// can be used.
OperatorRead() bool
// OperatorWrite determines if the state-changing Consul operator
// functions can be used.
OperatorWrite() bool
// PrepardQueryRead determines if a specific prepared query can be read
// to show its contents (this is not used for execution).
PreparedQueryRead(string) bool
// PreparedQueryWrite determines if a specific prepared query can be
// created, modified, or deleted.
PreparedQueryWrite(string) bool
// ServiceRead checks for permission to read a given service
ServiceRead(string) bool
// ServiceWrite checks for permission to create or update a given
// service
ServiceWrite(string) bool
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
// Snapshot checks for permission to take and restore snapshots.
Snapshot() bool
2014-08-06 22:08:17 +00:00
}
// StaticACL is used to implement a base ACL policy. It either
// allows or denies all requests. This can be used as a parent
// ACL to act in a blacklist or whitelist mode.
type StaticACL struct {
allowManage bool
2014-08-06 22:08:17 +00:00
defaultAllow bool
}
func (s *StaticACL) ACLList() bool {
return s.allowManage
2014-08-06 22:08:17 +00:00
}
func (s *StaticACL) ACLModify() bool {
return s.allowManage
2014-08-06 22:08:17 +00:00
}
func (s *StaticACL) EventRead(string) bool {
return s.defaultAllow
}
func (s *StaticACL) EventWrite(string) bool {
2014-12-01 03:33:46 +00:00
return s.defaultAllow
}
func (s *StaticACL) KeyRead(string) bool {
2014-12-01 03:33:46 +00:00
return s.defaultAllow
}
func (s *StaticACL) KeyWrite(string) bool {
2015-06-18 01:56:29 +00:00
return s.defaultAllow
}
func (s *StaticACL) KeyWritePrefix(string) bool {
2015-06-18 01:56:29 +00:00
return s.defaultAllow
}
func (s *StaticACL) KeyringRead() bool {
2015-07-07 00:28:09 +00:00
return s.defaultAllow
}
func (s *StaticACL) KeyringWrite() bool {
2015-07-07 00:28:09 +00:00
return s.defaultAllow
}
func (s *StaticACL) NodeRead(string) bool {
return s.defaultAllow
}
func (s *StaticACL) NodeWrite(string) bool {
return s.defaultAllow
}
func (s *StaticACL) OperatorRead() bool {
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
return s.defaultAllow
}
func (s *StaticACL) OperatorWrite() bool {
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
return s.defaultAllow
}
func (s *StaticACL) PreparedQueryRead(string) bool {
return s.defaultAllow
}
func (s *StaticACL) PreparedQueryWrite(string) bool {
return s.defaultAllow
}
func (s *StaticACL) ServiceRead(string) bool {
return s.defaultAllow
}
func (s *StaticACL) ServiceWrite(string) bool {
return s.defaultAllow
}
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
func (s *StaticACL) Snapshot() bool {
return s.allowManage
}
2014-08-06 22:08:17 +00:00
// AllowAll returns an ACL rule that allows all operations
func AllowAll() ACL {
return allowAll
}
// DenyAll returns an ACL rule that denies all operations
func DenyAll() ACL {
return denyAll
}
// ManageAll returns an ACL rule that can manage all resources
func ManageAll() ACL {
return manageAll
}
// RootACL returns a possible ACL if the ID matches a root policy
func RootACL(id string) ACL {
switch id {
case "allow":
return allowAll
case "deny":
return denyAll
case "manage":
return manageAll
default:
return nil
}
}
2014-08-06 22:08:17 +00:00
// PolicyACL is used to wrap a set of ACL policies to provide
// the ACL interface.
type PolicyACL struct {
// parent is used to resolve policy if we have
// no matching rule.
parent ACL
// keyRules contains the key policies
keyRules *radix.Tree
2014-12-01 03:33:46 +00:00
// nodeRules contains the node policies
nodeRules *radix.Tree
2014-12-01 03:33:46 +00:00
// serviceRules contains the service policies
2015-05-05 06:25:19 +00:00
serviceRules *radix.Tree
2015-06-18 01:56:29 +00:00
// eventRules contains the user event policies
eventRules *radix.Tree
2015-07-07 00:28:09 +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
// preparedQueryRules contains the prepared query policies
preparedQueryRules *radix.Tree
// keyringRule contains the keyring policies. The keyring has
2015-09-15 12:22:08 +00:00
// a very simple yes/no without prefix matching, so here we
2015-07-07 00:28:09 +00:00
// don't need to use a radix tree.
2015-07-07 16:45:38 +00:00
keyringRule string
// operatorRule contains the operator policies.
operatorRule string
2014-08-06 22:08:17 +00:00
}
// New is used to construct a policy based ACL from a set of policies
// and a parent policy to resolve missing cases.
func New(parent ACL, policy *Policy) (*PolicyACL, error) {
p := &PolicyACL{
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
parent: parent,
keyRules: radix.New(),
nodeRules: radix.New(),
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
serviceRules: radix.New(),
eventRules: radix.New(),
preparedQueryRules: radix.New(),
2014-08-06 22:08:17 +00:00
}
// Load the key policy
for _, kp := range policy.Keys {
p.keyRules.Insert(kp.Prefix, kp.Policy)
2014-08-06 22:08:17 +00:00
}
2014-12-01 03:33:46 +00:00
// Load the node policy
for _, np := range policy.Nodes {
p.nodeRules.Insert(np.Name, np.Policy)
}
2014-12-01 03:33:46 +00:00
// Load the service policy
for _, sp := range policy.Services {
2015-05-05 06:25:19 +00:00
p.serviceRules.Insert(sp.Name, sp.Policy)
2014-12-01 03:33:46 +00:00
}
2015-06-18 01:56:29 +00:00
// Load the event policy
for _, ep := range policy.Events {
p.eventRules.Insert(ep.Event, ep.Policy)
}
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
// Load the prepared query policy
for _, pq := range policy.PreparedQueries {
p.preparedQueryRules.Insert(pq.Prefix, pq.Policy)
}
2015-07-07 00:28:09 +00:00
// Load the keyring policy
2015-07-07 16:45:38 +00:00
p.keyringRule = policy.Keyring
2015-07-07 00:28:09 +00:00
// Load the operator policy
p.operatorRule = policy.Operator
2014-08-06 22:08:17 +00:00
return p, nil
}
// ACLList checks if listing of ACLs is allowed
func (p *PolicyACL) ACLList() bool {
return p.parent.ACLList()
}
// ACLModify checks if modification of ACLs is allowed
func (p *PolicyACL) ACLModify() bool {
return p.parent.ACLModify()
}
// Snapshot checks if taking and restoring snapshots is allowed.
func (p *PolicyACL) Snapshot() bool {
return p.parent.Snapshot()
}
// EventRead is used to determine if the policy allows for a
// specific user event to be read.
func (p *PolicyACL) EventRead(name string) bool {
// Longest-prefix match on event names
if _, rule, ok := p.eventRules.LongestPrefix(name); ok {
switch rule {
case PolicyRead, PolicyWrite:
return true
default:
return false
}
}
// Nothing matched, use parent
return p.parent.EventRead(name)
}
// EventWrite is used to determine if new events can be created
// (fired) by the policy.
func (p *PolicyACL) EventWrite(name string) bool {
// Longest-prefix match event names
if _, rule, ok := p.eventRules.LongestPrefix(name); ok {
return rule == PolicyWrite
}
// No match, use parent
return p.parent.EventWrite(name)
}
2014-08-06 22:08:17 +00:00
// KeyRead returns if a key is allowed to be read
func (p *PolicyACL) KeyRead(key string) bool {
// Look for a matching rule
_, rule, ok := p.keyRules.LongestPrefix(key)
2014-08-06 22:08:17 +00:00
if ok {
switch rule.(string) {
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
case PolicyRead, PolicyWrite:
return true
default:
return false
}
2014-08-06 22:08:17 +00:00
}
// No matching rule, use the parent.
return p.parent.KeyRead(key)
}
// KeyWrite returns if a key is allowed to be written
func (p *PolicyACL) KeyWrite(key string) bool {
// Look for a matching rule
_, rule, ok := p.keyRules.LongestPrefix(key)
2014-08-06 22:08:17 +00:00
if ok {
switch rule.(string) {
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
case PolicyWrite:
return true
default:
return false
}
2014-08-06 22:08:17 +00:00
}
// No matching rule, use the parent.
return p.parent.KeyWrite(key)
}
// KeyWritePrefix returns if a prefix is allowed to be written
func (p *PolicyACL) KeyWritePrefix(prefix string) bool {
// Look for a matching rule that denies
_, rule, ok := p.keyRules.LongestPrefix(prefix)
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 ok && rule.(string) != PolicyWrite {
return false
}
// Look if any of our children have a deny policy
deny := false
p.keyRules.WalkPrefix(prefix, func(path string, rule interface{}) bool {
// We have a rule to prevent a write in a sub-directory!
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 rule.(string) != PolicyWrite {
deny = true
return true
}
return false
})
// Deny the write if any sub-rules may be violated
if deny {
return false
}
// If we had a matching rule, done
if ok {
return true
}
// No matching rule, use the parent.
return p.parent.KeyWritePrefix(prefix)
}
// KeyringRead is used to determine if the keyring can be
// read by the current ACL token.
func (p *PolicyACL) KeyringRead() bool {
switch p.keyringRule {
case PolicyRead, PolicyWrite:
return true
case PolicyDeny:
return false
default:
return p.parent.KeyringRead()
2014-12-01 03:33:46 +00:00
}
}
// KeyringWrite determines if the keyring can be manipulated.
func (p *PolicyACL) KeyringWrite() bool {
if p.keyringRule == PolicyWrite {
return true
2014-12-01 03:33:46 +00:00
}
return p.parent.KeyringWrite()
2014-12-01 03:33:46 +00:00
}
// OperatorRead determines if the read-only operator functions are allowed.
func (p *PolicyACL) OperatorRead() bool {
switch p.operatorRule {
case PolicyRead, PolicyWrite:
return true
case PolicyDeny:
return false
default:
return p.parent.OperatorRead()
2015-06-18 01:56:29 +00:00
}
}
// NodeRead checks if reading (discovery) of a node is allowed
func (p *PolicyACL) NodeRead(name string) bool {
// Check for an exact rule or catch-all
_, rule, ok := p.nodeRules.LongestPrefix(name)
if ok {
switch rule {
case PolicyRead, PolicyWrite:
return true
default:
return false
}
}
// No matching rule, use the parent.
return p.parent.NodeRead(name)
}
// NodeWrite checks if writing (registering) a node is allowed
func (p *PolicyACL) NodeWrite(name string) bool {
// Check for an exact rule or catch-all
_, rule, ok := p.nodeRules.LongestPrefix(name)
if ok {
switch rule {
case PolicyWrite:
return true
default:
return false
}
}
// No matching rule, use the parent.
return p.parent.NodeWrite(name)
}
// OperatorWrite determines if the state-changing operator functions are
// allowed.
func (p *PolicyACL) OperatorWrite() bool {
if p.operatorRule == PolicyWrite {
return true
2015-06-18 01:56:29 +00:00
}
return p.parent.OperatorWrite()
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
// PreparedQueryRead checks if reading (listing) of a prepared query is
// allowed - this isn't execution, just listing its contents.
func (p *PolicyACL) PreparedQueryRead(prefix string) bool {
// Check for an exact rule or catch-all
_, rule, ok := p.preparedQueryRules.LongestPrefix(prefix)
if ok {
switch rule {
case PolicyRead, PolicyWrite:
return true
default:
return false
}
}
// No matching rule, use the parent.
return p.parent.PreparedQueryRead(prefix)
}
// PreparedQueryWrite checks if writing (creating, updating, or deleting) of a
// prepared query is allowed.
func (p *PolicyACL) PreparedQueryWrite(prefix string) bool {
// Check for an exact rule or catch-all
_, rule, ok := p.preparedQueryRules.LongestPrefix(prefix)
if ok {
switch rule {
case PolicyWrite:
return true
default:
return false
}
}
// No matching rule, use the parent.
return p.parent.PreparedQueryWrite(prefix)
}
// ServiceRead checks if reading (discovery) of a service is allowed
func (p *PolicyACL) ServiceRead(name string) bool {
// Check for an exact rule or catch-all
_, rule, ok := p.serviceRules.LongestPrefix(name)
2015-07-07 00:28:09 +00:00
if ok {
switch rule {
case PolicyRead, PolicyWrite:
return true
default:
return false
}
}
// No matching rule, use the parent.
return p.parent.ServiceRead(name)
}
// ServiceWrite checks if writing (registering) a service is allowed
func (p *PolicyACL) ServiceWrite(name string) bool {
// Check for an exact rule or catch-all
_, rule, ok := p.serviceRules.LongestPrefix(name)
if ok {
switch rule {
case PolicyWrite:
return true
default:
return false
}
}
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
// No matching rule, use the parent.
return p.parent.ServiceWrite(name)
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
}