ACL Rules

1.4.0 and later: This document only applies in Consul versions 1.4.0 and later. The documentation for the legacy ACL system is here

Consul provides an optional Access Control List (ACL) system which can be used to control access to data and APIs. To learn more about Consul’s ACL review the ACL system documentation

A core part of the ACL system is the rule language, which is used to describe the policy that must be enforced. There are two types of rules: prefix based rules and exact matching rules.

Rule Specification

Rules are composed of a resource, a segment (for some resource areas) and a policy disposition. The general structure of a rule is:

  1. <resource> "<segment>" {
  2. policy = "<policy disposition>"
  3. }

Segmented resource areas allow operators to more finely control access to those resources. Note that not all resource areas are segmented such as the keyring, operator, and acl resources. For those rules they would look like:

  1. <resource> = "<policy disposition>"

Policies can have several control levels:

  • read: allow the resource to be read but not modified.
  • write: allow the resource to be read and modified.
  • deny: do not allow the resource to be read or modified.
  • list: allows access to all the keys under a segment in the Consul KV. Note, this policy can only be used with the key_prefix resource and acl.enable_key_list_policy must be set to true.

When using prefix-based rules, the most specific prefix match determines the action. This allows for flexible rules like an empty prefix to allow read-only access to all resources, along with some specific prefixes that allow write access or that are denied all access. Exact matching rules will only apply to the exact resource specified. The order of precedence for matching rules are, DENY has priority over WRITE or READ and WRITE has priority over READ.

We make use of the HashiCorp Configuration Language (HCL) to specify rules. This language is human readable and interoperable with JSON making it easy to machine-generate. Rules can make use of one or more policies.

Specification in the HCL format looks like:

  1. # These control access to the key/value store.
  2. key_prefix "" {
  3. policy = "read"
  4. }
  5. key_prefix "foo/" {
  6. policy = "write"
  7. }
  8. key_prefix "foo/private/" {
  9. policy = "deny"
  10. }
  11. # Or for exact key matches
  12. key "foo/bar/secret" {
  13. policy = "deny"
  14. }
  15. # This controls access to cluster-wide Consul operator information.
  16. operator = "read"

This is equivalent to the following JSON input:

  1. {
  2. "key_prefix": {
  3. "": {
  4. "policy": "read"
  5. },
  6. "foo/": {
  7. "policy": "write"
  8. },
  9. "foo/private/": {
  10. "policy": "deny"
  11. }
  12. },
  13. "key": {
  14. "foo/bar/secret": {
  15. "policy": "deny"
  16. }
  17. },
  18. "operator": "read"
  19. }

The ACL API allows either HCL or JSON to be used to define the content of the rules section of a policy.

Here’s a sample request using the HCL form:

  1. $ curl \
  2. --request PUT \
  3. --data \
  4. '{
  5. "Name": "my-app-policy",
  6. "Rules": "key \"\" { policy = \"read\" } key \"foo/\" { policy = \"write\" } key \"foo/private/\" { policy = \"deny\" } operator = \"read\""
  7. }' http://127.0.0.1:8500/v1/acl/policy?token=<token with ACL "write">

Here’s an equivalent request using the JSON form:

  1. $ curl \
  2. --request PUT \
  3. --data \
  4. '{
  5. "Name": "my-app-policy",
  6. "Rules": "{\"key\":{\"\":{\"policy\":\"read\"},\"foo/\":{\"policy\":\"write\"},\"foo/private\":{\"policy\":\"deny\"}},\"operator\":\"read\"}"
  7. }' http://127.0.0.1:8500/v1/acl/policy?token=<management token>

On success, the Policy is returned:

  1. {
  2. "CreateIndex": 7,
  3. "Hash": "UMG6QEbV40Gs7Cgi6l/ZjYWUwRS0pIxxusFKyKOt8qI=",
  4. "ID": "5f423562-aca1-53c3-e121-cb0eb2ea1cd3",
  5. "ModifyIndex": 7,
  6. "Name": "my-app-policy",
  7. "Rules": "key \"\" { policy = \"read\" } key \"foo/\" { policy = \"write\" } key \"foo/private/\" { policy = \"deny\" } operator = \"read\""
  8. }

The created policy can now be specified either by name or by ID when creating a token. This will grant the rules provided to the bearer of that token.

Below is a breakdown of each rule type.

ACL Resource Rules

The acl resource controls access to ACL operations in the ACL API.

ACL rules look like this:

  1. acl = "write"

There is only one acl rule allowed per policy and its value is set to one of the policy dispositions. In the example above ACLs may be read or written including discovering any token’s secret ID. Snapshotting also requires acl = "write" permissions due to the fact that all the token secrets are contained within the snapshot.

Agent Rules

The agent and agent_prefix resources control access to the utility operations in the Agent API, such as join and leave. All of the catalog-related operations are covered by the node or node_prefix and service or service_prefix policies instead.

Agent rules look like this:

  1. agent_prefix "" {
  2. policy = "read"
  3. }
  4. agent "foo" {
  5. policy = "write"
  6. }
  7. agent_prefix "bar" {
  8. policy = "deny"
  9. }

Agent rules are keyed by the node name they apply to. In the example above the rules allow read-only access to any node name by using the empty prefix, read-write access to the node with the exact name foo, and denies all access to any node name that starts with bar.

Since Agent API utility operations may be required before an agent is joined to a cluster, or during an outage of the Consul servers or ACL datacenter, a special token may be configured with acl.tokens.agent_master to allow write access to these operations even if no ACL resolution capability is available.

Event Rules

The event and event_prefix resources control access to event operations in the Event API, such as firing events and listing events.

Event rules look like this:

  1. event_prefix "" {
  2. policy = "read"
  3. }
  4. event "deploy" {
  5. policy = "write"
  6. }

Event rules are segmented by the event name they apply to. In the example above, the rules allow read-only access to any event, and firing of the “deploy” event.

The consul exec command uses events with the “_rexec” prefix during operation, so to enable this feature in a Consul environment with ACLs enabled, you will need to give agents a token with access to this event prefix, in addition to configuring disable_remote_exec to false.

Key/Value Rules

The key and key_prefix resources control access to key/value store operations in the KV API. Key rules look like this:

  1. key_prefix "" {
  2. policy = "read"
  3. }
  4. key "foo" {
  5. policy = "write"
  6. }
  7. key "bar" {
  8. policy = "deny"
  9. }

Key rules are segmented by the key name they apply to. In the example above, the rules allow read-only access to any key name with the empty prefix rule, allow read-write access to the “foo” key, and deny access to the “bar” key.

List Policy for Keys

Consul 1.0 introduces a new list policy for keys that is only enforced when opted in via the boolean config param “acl.enable_key_list_policy”. list controls access to recursively list entries and keys, and enables more fine grained policies. With “acl.enable_key_list_policy”, recursive reads via the KV API with an invalid token result in a 403. Example:

  1. key_prefix "" {
  2. policy = "deny"
  3. }
  4. key_prefix "bar" {
  5. policy = "list"
  6. }
  7. key_prefix "baz" {
  8. policy = "read"
  9. }

In the example above, the rules allow reading the key “baz”, and only allow recursive reads on the prefix “bar”.

A token with write access on a prefix also has list access. A token with list access on a prefix also has read access on all its suffixes.

Sentinel Integration

Enterprise

Consul Enterprise supports additional optional fields for key write policies for Sentinel integration. An example key rule with a Sentinel code policy looks like this:

  1. key "foo" {
  2. policy = "write"
  3. sentinel {
  4. code = <<EOF
  5. import "strings"
  6. main = rule { strings.has_suffix(value, "bar") }
  7. EOF
  8. enforcementlevel = "hard-mandatory"
  9. }
  10. }

For more detailed information, see the Consul Sentinel documentation.

Keyring Rules

The keyring resource controls access to keyring operations in the Keyring API.

Keyring rules look like this:

  1. keyring = "write"

There’s only one keyring policy allowed per rule set, and its value is set to one of the policy dispositions. In the example above, the keyring may be read and updated.

Node Rules

The node and node_prefix resources controls node-level registration and read access to the Catalog API, service discovery with the Health API, and filters results in Agent API operations like fetching the list of cluster members.

Node rules look like this:

  1. node_prefix "" {
  2. policy = "read"
  3. }
  4. node "app" {
  5. policy = "write"
  6. }
  7. node "admin" {
  8. policy = "deny"
  9. }

Node rules are segmented by the node name they apply to. In the example above, the rules allow read-only access to any node name with the empty prefix, allow read-write access to the “app” node, and deny all access to the “admin” node.

Agents need to be configured with an acl.tokens.agent with at least “write” privileges to their own node name in order to register their information with the catalog, such as node metadata and tagged addresses. If this is configured incorrectly, the agent will print an error to the console when it tries to sync its state with the catalog.

Consul’s DNS interface is also affected by restrictions on node rules. If the acl.token.default used by the agent does not have “read” access to a given node, then the DNS interface will return no records when queried for it.

When reading from the catalog or retrieving information from the health endpoints, node rules are used to filter the results of the query. This allows for configurations where a token has access to a given service name, but only on an allowed subset of node names.

Node rules come into play when using the Agent API to register node-level checks. The agent will check tokens locally as a check is registered, and Consul also performs periodic anti-entropy syncs, which may require an ACL token to complete. To accommodate this, Consul provides two methods of configuring ACL tokens to use for registration events:

  1. Using the acl.tokens.default configuration directive. This allows a single token to be configured globally and used during all check registration operations.
  2. Providing an ACL token with service and check definitions at registration time. This allows for greater flexibility and enables the use of multiple tokens on the same agent. Examples of what this looks like are available for both services and checks. Tokens may also be passed to the HTTP API for operations that require them.

In addition to ACLs, in Consul 0.9.0 and later, the agent must be configured with enable_script_checks set to true in order to enable script checks.

Operator Rules

The operator resource controls access to cluster-level operations in the Operator API, other than the Keyring API.

Operator rules look like this:

  1. operator = "read"

There’s only one operator rule allowed per rule set, and its value is set to one of the policy dispositions. In the example above, the token could be used to query the operator endpoints for diagnostic purposes but not make any changes.

Prepared Query Rules

The query and query_prefix resources control access to create, update, and delete prepared queries in the Prepared Query API. Executing queries is subject to node/node_prefix and service/service_prefix policies, as will be explained below.

Query rules look like this:

  1. query_prefix "" {
  2. policy = "read"
  3. }
  4. query "foo" {
  5. policy = "write"
  6. }

Query rules are segmented by the query name they apply to. In the example above, the rules allow read-only access to any query name with the empty prefix, and allow read-write access to the query named “foo”. This allows control of the query namespace to be delegated based on ACLs.

There are a few variations when using ACLs with prepared queries, each of which uses ACLs in one of two ways: open, protected by unguessable IDs or closed, managed by ACL policies. These variations are covered here, with examples:

  • Static queries with no Name defined are not controlled by any ACL policies. These types of queries are meant to be ephemeral and not shared to untrusted clients, and they are only reachable if the prepared query ID is known. Since these IDs are generated using the same random ID scheme as ACL Tokens, it is infeasible to guess them. When listing all prepared queries, only a management token will be able to see these types, though clients can read instances for which they have an ID. An example use for this type is a query built by a startup script, tied to a session, and written to a configuration file for a process to use via DNS.

  • Static queries with a Name defined are controlled by the query and query_prefix ACL resources. Clients are required to have an ACL token with permissions on to access that query name. Clients can list or read queries for which they have “read” access based on their prefix, and similar they can update any queries for which they have “write” access. An example use for this type is a query with a well-known name (eg. prod-master-customer-db) that is used and known by many clients to provide geo-failover behavior for a database.

  • Template queries queries work like static queries with a Name defined, except that a catch-all template with an empty Name requires an ACL token that can write to any query prefix.

When prepared queries are executed via DNS lookups or HTTP requests, the ACL checks are run against the service being queried, similar to how ACLs work with other service lookups. There are several ways the ACL token is selected for this check:

  • If an ACL Token was captured when the prepared query was defined, it will be used to perform the service lookup. This allows queries to be executed by clients with lesser or even no ACL Token, so this should be used with care.

  • If no ACL Token was captured, then the client’s ACL Token will be used to perform the service lookup.

  • If no ACL Token was captured and the client has no ACL Token, then the anonymous token will be used to perform the service lookup.

In the common case, the ACL Token of the invoker is used to test the ability to look up a service. If a Token was specified when the prepared query was created, the behavior changes and now the captured ACL Token set by the definer of the query is used when looking up a service.

Capturing ACL Tokens is analogous to PostgreSQL’s SECURITY DEFINER attribute which can be set on functions, and using the client’s ACL Token is similar to the complementary SECURITY INVOKER attribute.

Prepared queries were originally introduced in Consul 0.6.0, and ACL behavior remained unchanged through version 0.6.3, but was then changed to allow better management of the prepared query namespace.

These differences are outlined in the table below:

OperationVersion <= 0.6.3Version > 0.6.3
Create static query without NameThe ACL Token used to create the prepared query is checked to make sure it can access the service being queried. This token is captured as the Token to use when executing the prepared query.No ACL policies are used as long as no Name is defined. No Token is captured by default unless specifically supplied by the client when creating the query.
Create static query with NameThe ACL Token used to create the prepared query is checked to make sure it can access the service being queried. This token is captured as the Token to use when executing the prepared query.The client token’s query ACL policy is used to determine if the client is allowed to register a query for the given Name. No Token is captured by default unless specifically supplied by the client when creating the query.
Manage static query without NameThe ACL Token used to create the query or a token with management privileges must be supplied in order to perform these operations.Any client with the ID of the query can perform these operations.
Manage static query with a NameThe ACL token used to create the query or a token with management privileges must be supplied in order to perform these operations.Similar to create, the client token’s query ACL policy is used to determine if these operations are allowed.
List queriesA token with management privileges is required to list any queries.The client token’s query ACL policy is used to determine which queries they can see. Only tokens with management privileges can see prepared queries without Name.
Execute querySince a Token is always captured when a query is created, that is used to check access to the service being queried. Any token supplied by the client is ignored.The captured token, client’s token, or anonymous token is used to filter the results, as described above.

Service Rules

The service and service_prefix resources control service-level registration and read access to the Catalog API and service discovery with the Health API.

Service rules look like this:

  1. service_prefix "" {
  2. policy = "read"
  3. }
  4. service "app" {
  5. policy = "write"
  6. }
  7. service "admin" {
  8. policy = "deny"
  9. }

Service rules are segmented by the service name they apply to. In the example above, the rules allow read-only access to any service name with the empty prefix, allow read-write access to the “app” service, and deny all access to the “admin” service.

Consul’s DNS interface is affected by restrictions on service rules. If the acl.tokens.default used by the agent does not have “read” access to a given service, then the DNS interface will return no records when queried for it.

When reading from the catalog or retrieving information from the health endpoints, service rules are used to filter the results of the query.

Service rules come into play when using the Agent API to register services or checks. The agent will check tokens locally as a service or check is registered, and Consul also performs periodic anti-entropy syncs, which may require an ACL token to complete. To accommodate this, Consul provides two methods of configuring ACL tokens to use for registration events:

  1. Using the acl.tokens.default configuration directive. This allows a single token to be configured globally and used during all service and check registration operations.
  2. Providing an ACL token with service and check definitions at registration time. This allows for greater flexibility and enables the use of multiple tokens on the same agent. Examples of what this looks like are available for both services and checks. Tokens may also be passed to the HTTP API for operations that require them. Note: all tokens passed to an agent are persisted on local disk to allow recovery from restarts. See -data-dir flag documentation for notes on securing access.

In addition to ACLs, in Consul 0.9.0 and later, the agent must be configured with enable_script_checks or enable_local_script_checks set to true in order to enable script checks.

Note: Intention privileges are managed with service rules.

Session Rules

The session and session_prefix resources controls access to Session API operations.

Session rules look like this:

  1. session_prefix "" {
  2. policy = "read"
  3. }
  4. session "app" {
  5. policy = "write"
  6. }
  7. session "admin" {
  8. policy = "deny"
  9. }

Session rules are segmented by the node name they apply to. In the example above, the rules allow read-only access to sessions on node name with the empty prefix, allow creating sessions on the node named “app”, and deny all access to any sessions on the “admin” node.

Namespace Rules

Enterprise

Consul Enterprise 1.7.0 adds support for namespacing many Consul resources. ACL rules themselves can then be defined to only to apply to specific namespaces.

A Namespace specific rule would look like this:

  1. namespace_prefix "" {
  2. # grant service:read for all services in all namespaces
  3. service_prefix "" {
  4. policy = "read"
  5. }
  6. # grant node:read for all nodes in all namespaces
  7. node_prefix "" {
  8. policy = "read"
  9. }
  10. }
  11. namespace "foo" {
  12. # grants permission to manage ACLs only for the foo namespace
  13. acl = "write"
  14. # grants write permissions to the KV for namespace foo
  15. key_prefix "" {
  16. policy = "write"
  17. }
  18. # grants write permissions for sessions for namespace foo
  19. session_prefix "" {
  20. policy = "write"
  21. }
  22. # grants service:write for all services in the foo namespace
  23. service_prefix "" {
  24. policy = "write"
  25. }
  26. # grants node:read for all nodes
  27. node_prefix "" {
  28. policy = "read"
  29. }
  30. }

Note, when a rule is defined in any user created namespace, the following restrictions apply.

  1. operator rules are not allowed.
  2. event rules are not allowed.
  3. keyring rules are not allowed.
  4. query rules are not allowed.
  5. node rules that attempt to grant write privileges are not allowed.

These restrictions do not apply to the default namespace created by Consul. In general all of the above are permissions that only an operator should have and thus granting these permissions can only be done within the default namespace.

Implicit namespacing: Rules and policies created within a namespace will inherit the namespace configuration. This means that rules and policies will be implicitly namespaced and do not need additional configuration. The restrictions outlined above will apply to these rules and policies. Additionally, rules and policies within a specific namespace are prevented from accessing resources in another namespace.