» Prepared Query HTTP Endpoint

The /query endpoints create, update, destroy, and execute prepared queries.

Prepared queries allow you to register a complex service query and then execute it later via its ID or name to get a set of healthy nodes that provide a given service. This is particularly useful in combination with Consul's DNS Interface as it allows for much richer queries than would be possible given the limited entry points exposed by DNS.

See the Geo Failover Guide for details and examples for using prepared queries to implement geo failover for services.

See the prepared query rules section of the agent ACL documentation for more details about how prepared queries work with Consul's ACL system.

» Prepared Query Templates

Consul 0.6.4 and later support prepared query templates. These are created similar to static queries, except with some additional fields and features. Here is an example prepared query template:

{
  "Template": {
    "Type": "name_prefix_match",
    "Regexp": "^geo-db-(.*?)-([^\\-]+?)$",
    "RemoveEmptyTags": false
  }
}

The Template structure configures a prepared query as a template instead of a static query. It has two fields:

  • Type is the query type, which must be name_prefix_match. This means that the template will apply to any query lookup with a name whose prefix matches the Name field of the template. In this example, any query for geo-db will match this query. Query templates are resolved using a longest prefix match, so it's possible to have high-level templates that are overridden for specific services. Static queries are always resolved first, so they can also override templates.

  • Regexp is an optional regular expression which is used to extract fields from the entire name, once this template is selected. In this example, the regular expression takes the first item after the "-" as the database name and everything else after as a tag. See the RE2 reference for syntax of this regular expression.

  • RemoveEmptyTags is optional, and if set to true, will cause the Tags list inside the Service structure to be stripped of any empty strings. This defaults to false, meaning that empty strings will remain in the list. This is useful when interpolating into tags in a way where the tag is optional, and where searching for an empty tag would yield no results from the query.

All other fields of the query have the same meanings as for a static query, except that several interpolation variables are available to dynamically populate the query before it is executed. All of the string fields inside the Service structure are interpolated, with the following variables available:

  • ${name.full} has the entire name that was queried. For example, a DNS lookup for geo-db-customer-primary.query.consul in the example above would set this variable to geo-db-customer-primary.

  • ${name.prefix} has the prefix that matched. This would always be geo-db for the example above.

  • ${name.suffix} has the suffix after the prefix. For example, a DNS lookup for geo-db-customer-primary.query.consul in the example above would set this variable to -customer-primary.

  • ${match(N)} returns the regular expression match at the given index N. The 0 index will have the entire match, and >0 will have the results of each match group. For example, a DNS lookup for geo-db-customer-primary.query.consul in the example above with a Regexp field set to ^geo-db-(.*?)-([^\-]+?)$ would return geo-db-customer-primary for ${match(0)}, customer for ${match(1)}, and primary for ${match(2)}. If the regular expression doesn't match, or an invalid index is given, then ${match(N)} will return an empty string.

  • ${agent.segment} has the network segment (Enterprise-only) of the agent that initiated the query. This can be used with the NodeMeta field to limit the results of a query to service instances within its own network segment:

    {
  "Name": "",
  "Template": {
    "Type": "name_prefix_match"
  },
  "Service": {
    "Service": "${name.full}",
    "NodeMeta": {"consul-network-segment": "${agent.segment}"}
  }
}

    This will map all names of the form <service>.query.consul over DNS to a query that will select an instance of the service in the agent's own network segment.

Using templates, it is possible to apply prepared query behaviors to many services with a single template. Here's an example template that matches any query and applies a failover policy to it:

{
  "Name": "",
  "Template": {
    "Type": "name_prefix_match"
  },
  "Service": {
    "Service": "${name.full}",
    "Failover": {
      "NearestN": 3
    }
  }
}

This will match any lookup for *.query.consul and will attempt to find the service locally, and otherwise attempt to find that service in the next three closest datacenters. If ACLs are enabled, a catch-all template like this with an empty Name requires an ACL token that can write to any query prefix. Also, only a single catch-all template can be registered at any time.

» Create Prepared Query

This endpoint creates a new prepared query and returns its ID if it is created successfully.

Method Path Produces
POST /query application/json

The table below shows this endpoint's support for blocking queries, consistency modes, agent caching, and required ACLs.

Blocking Queries Consistency Modes Agent Caching ACL Required
NO none none query:write

» Parameters

  • dc (string: "") - Specifies the datacenter to query. This will default to the datacenter of the agent being queried. This is specified as part of the URL as a query parameter.

  • Name (string: "") - Specifies an optional friendly name that can be used to execute a query instead of using its ID.

  • Session (string: "") - Specifies the ID of an existing session. This provides a way to automatically remove a prepared query when the given session is invalidated. If not given the prepared query must be manually removed when no longer needed.

  • Token (string: "") - Specifies the ACL token to use each time the query is executed. This allows queries to be executed by clients with lesser or even no ACL Token, so this should be used with care. The token itself can only be seen by clients with a management token. If the Token field is left blank or omitted, the client's ACL Token will be used to determine if they have access to the service being queried. If the client does not supply an ACL Token, the anonymous token will be used.

  • Service (Service: <required>) - Specifies the structure to define the query's behavior.

    • Service (string: <required>) - Specifies the name of the service to query.
    • Failover contains two fields, both of which are optional, and determine what happens if no healthy nodes are available in the local datacenter when the query is executed. It allows the use of nodes in other datacenters with very little configuration.
      • NearestN (int: 0) - Specifies that the query will be forwarded to up to NearestN other datacenters based on their estimated network round trip time using Network Coordinates from the WAN gossip pool. The median round trip time from the server handling the query to the servers in the remote datacenter is used to determine the priority.
      • Datacenters (array<string>: nil) - Specifies a fixed list of remote datacenters to forward the query to if there are no healthy nodes in the local datacenter. Datacenters are queried in the order given in the list. If this option is combined with NearestN, then the NearestN queries will be performed first, followed by the list given by Datacenters. A given datacenter will only be queried one time during a failover, even if it is selected by both NearestN and is listed in Datacenters.
    • IgnoreCheckIDs (array<string>: nil) - Specifies a list of check IDs that should be ignored when filtering unhealthy instances. This is mostly useful in an emergency or as a temporary measure when a health check is found to be unreliable. Being able to ignore it in centrally-defined queries can be simpler than de-registering the check as an interim solution until the check can be fixed.
    • OnlyPassing (bool: false) - Specifies the behavior of the query's health check filtering. If this is set to false, the results will include nodes with checks in the passing as well as the warning states. If this is set to true, only nodes with checks in the passing state will be returned.
    • Near (string: "") - Specifies a node to sort near based on distance sorting using Network Coordinates. The nearest instance to the specified node will be returned first, and subsequent nodes in the response will be sorted in ascending order of estimated round-trip times. If the node given does not exist, the nodes in the response will be shuffled. If unspecified, the response will be shuffled by default.
      • _agent - Returns results nearest the agent servicing the request.
      • _ip - Returns results nearest to the node associated with the source IP where the query was executed from. For HTTP the source IP is the remote peer's IP address or the value of the X-Forwarded-For header with the header taking precedence. For DNS the source IP is the remote peer's IP address or the value of the ENDS client IP with the EDNS client IP taking precedence.
    • Tags (array<string>: nil) - Specifies a list of service tags to filter the query results. For a service to pass the tag filter it must have all of the required tags, and none of the excluded tags (prefixed with !).
    • NodeMeta (map<string|string>: nil) - Specifies a list of user-defined key/value pairs that will be used for filtering the query results to nodes with the given metadata values present.
    • ServiceMeta (map<string|string>: nil) - Specifies a list of user-defined key/value pairs that will be used for filtering the query results to services with the given metadata values present.
    • Connect (bool: false) - If true, only Connect-capable services for the specified service name will be returned. This includes both natively integrated services and proxies. For proxies, the proxy name may not match Service, because the proxy destination will. Any constrains beyond the service name such as Near, Tags, and NodeMeta are applied to Connect-capable service.

  • DNS (DNS: nil) - Specifies DNS configuration

    • TTL (string: "") - Specifies the TTL duration when query results are served over DNS. If this is specified, it will take precedence over any Consul agent-specific configuration.

» Sample Payload 

{
  "Name": "my-query",
  "Session": "adf4238a-882b-9ddc-4a9d-5b6758e4159e",
  "Token": "",
  "Service": {
    "Service": "redis",
    "Failover": {
      "NearestN": 3,
      "Datacenters": ["dc1", "dc2"]
    },
    "Near": "node1",
    "OnlyPassing": false,
    "Tags": ["primary", "!experimental"],
    "NodeMeta": {"instance_type": "m3.large"},
    "ServiceMeta": {"environment": "production"}
  },
  "DNS": {
    "TTL": "10s"
  }
}

» Sample Request 

$ curl \
    --request POST \
    --data @payload.json \
    http://127.0.0.1:8500/v1/query

» Sample Response 

{
  "ID": "8f246b77-f3e1-ff88-5b48-8ec93abf3e05"
}

» Read Prepared Query

This endpoint returns a list of all prepared queries.

Method Path Produces
GET /query application/json

The table below shows this endpoint's support for blocking queries, consistency modes, agent caching, and required ACLs.

Blocking Queries Consistency Modes Agent Caching ACL Required
NO none none query:read

» Parameters

  • dc (string: "") - Specifies the datacenter to query. This will default to the datacenter of the agent being queried. This is specified as part of the URL as a query parameter.

» Sample Request 

$ curl \
    http://127.0.0.1:8500/v1/query

» Sample Response 

[
  {
    "ID": "8f246b77-f3e1-ff88-5b48-8ec93abf3e05",
    "Name": "my-query",
    "Session": "adf4238a-882b-9ddc-4a9d-5b6758e4159e",
    "Token": "<hidden>",
    "Service": {
      "Service": "redis",
      "Failover": {
        "NearestN": 3,
        "Datacenters": ["dc1", "dc2"]
      },
      "OnlyPassing": false,
      "Tags": ["primary", "!experimental"],
      "NodeMeta": {"instance_type": "m3.large"},
      "ServiceMeta": {"environment": "production"}
    },
    "DNS": {
      "TTL": "10s"
    },
    "RaftIndex": {
      "CreateIndex": 23,
      "ModifyIndex": 42
    }
  }
]

» Update Prepared Query

This endpoint updates an existing prepared query. If no query exists by the given ID, an error is returned.

Method Path Produces
PUT /query/:uuid application/json

The table below shows this endpoint's support for blocking queries, consistency modes, agent caching, and required ACLs.

Blocking Queries Consistency Modes Agent Caching ACL Required
NO none none query:write

» Parameters

  • uuid (string: <required>) - Specifies the UUID of the query to update. This is required and is specified as part of the URL path.

  • dc (string: "") - Specifies the datacenter to query. This will default to the datacenter of the agent being queried. This is specified as part of the URL as a query parameter.

The body is the same as is used to create a prepared query. Please see above for more information.

» Sample Request 

$ curl \
    --request PUT \
    --data @payload.json \
    http://127.0.0.1:8500/v1/query/8f246b77-f3e1-ff88-5b48-8ec93abf3e05

» Read Prepared Query

This endpoint reads an existing prepared query. If no query exists by the given ID, an error is returned.

Method Path Produces
GET /query/:uuid application/json

The table below shows this endpoint's support for blocking queries, consistency modes, agent caching, and required ACLs.

Blocking Queries Consistency Modes Agent Caching ACL Required
NO none none query:read

» Parameters

  • uuid (string: <required>) - Specifies the UUID of the query to read. This is required and is specified as part of the URL path.

  • dc (string: "") - Specifies the datacenter to query. This will default to the datacenter of the agent being queried. This is specified as part of the URL as a query parameter.

» Sample Request 

$ curl \
  http://127.0.0.1:8500/v1/query/8f246b77-f3e1-ff88-5b48-8ec93abf3e05

» Sample Response

The returned response is the same as the list of prepared queries above, only with a single item present.

» Delete Prepared Query

This endpoint deletes an existing prepared query. If no query exists by the given ID, an error is returned.

Method Path Produces
DELETE /query/:uuid application/json

The table below shows this endpoint's support for blocking queries, consistency modes, agent caching, and required ACLs.

Blocking Queries Consistency Modes Agent Caching ACL Required
NO none none query:write

» Parameters

  • uuid (string: <required>) - Specifies the UUID of the query to delete. This is required and is specified as part of the URL path.

  • dc (string: "") - Specifies the datacenter to query. This will default to the datacenter of the agent being queried. This is specified as part of the URL as a query parameter.

» Sample Request 

$ curl \
    --request DELETE \
    http://127.0.0.1:8500/v1/query/8f246b77-f3e1-ff88-5b48-8ec93abf3e05

» Execute Prepared Query

This endpoint executes an existing prepared query. If no query exists by the given ID, an error is returned.

Method Path Produces
GET /query/:uuid/execute application/json

The table below shows this endpoint's support for blocking queries, consistency modes, agent caching, and required ACLs.

Blocking Queries Consistency Modes Agent Caching ACL Required
NO none simple depends1

1 If an ACL Token was bound to the query when it was defined then it will be used when executing the request. Otherwise, the client's supplied ACL Token will be used.

» Parameters

  • uuid (string: <required>) - Specifies the UUID of the query to execute. This is required and is specified as part of the URL path. This can also be the name of an existing prepared query, or a name that matches a prefix name for a prepared query template.

  • dc (string: "") - Specifies the datacenter to query. This will default to the datacenter of the agent being queried. This is specified as part of the URL as a query parameter.

  • near (string: "") - Specifies to sort the resulting list in ascending order based on the estimated round trip time from that node. Passing ?near=_agent will use the agent's node for the sort. Passing ?near=_ip will use the source IP of the request or the value of the X-Forwarded-For header to lookup the node to use for the sort. If this is not present, the default behavior will shuffle the nodes randomly each time the query is executed.

  • limit (int: 0) - Limit the size of the list to the given number of nodes. This is applied after any sorting or shuffling.

  • connect (bool: false) - If true, limit results to nodes that are Connect-capable only. This can also be specified directly on the template itself to force all executions of a query to be Connect-only. See the template documentation for more information.

» Sample Request 

$ curl \
    http://127.0.0.1:8500/v1/query/8f246b77-f3e1-ff88-5b48-8ec93abf3e05/execute?near=_agent

» Sample Response 

{
  "Service": "redis",
  "Nodes": [
    {
      "Node": {
        "ID": "40e4a748-2192-161a-0510-9bf59fe950b5",
        "Node": "foobar",
        "Address": "10.1.10.12",
        "Datacenter": "dc1",
        "TaggedAddresses": {
          "lan": "10.1.10.12",
          "wan": "10.1.10.12"
        },
        "NodeMeta": {"instance_type": "m3.large"}
      },
      "Service": {
        "ID": "redis",
        "Service": "redis",
        "Tags": null,
        "Meta": {"redis_version": "4.0"},
        "Port": 8000
      },
      "Checks": [
        {
          "Node": "foobar",
          "CheckID": "service:redis",
          "Name": "Service 'redis' check",
          "Status": "passing",
          "Notes": "",
          "Output": "",
          "ServiceID": "redis",
          "ServiceName": "redis"
        },
        {
          "Node": "foobar",
          "CheckID": "serfHealth",
          "Name": "Serf Health Status",
          "Status": "passing",
          "Notes": "",
          "Output": "",
          "ServiceID": "",
          "ServiceName": ""
        }
      ],
    "DNS": {
      "TTL": "10s"
    },
    "Datacenter": "dc3",
    "Failovers": 2
  }]
}

  • Nodes contains the list of healthy nodes providing the given service, as specified by the constraints of the prepared query.

  • Service has the service name that the query was selecting. This is useful for context in case an empty list of nodes is returned.

  • DNS has information used when serving the results over DNS. This is just a copy of the structure given when the prepared query was created.

  • Datacenter has the datacenter that ultimately provided the list of nodes and Failovers has the number of remote datacenters that were queried while executing the query. This provides some insight into where the data came from. This will be zero during non-failover operations where there were healthy nodes found in the local datacenter.

» Explain Prepared Query

This endpoint generates a fully-rendered query for a given name, post interpolation.

Method Path Produces
GET /query/:uuid/explain application/json

The table below shows this endpoint's support for blocking queries, consistency modes, agent caching, and required ACLs.

Blocking Queries Consistency Modes Agent Caching ACL Required
NO none none query:read

» Parameters

  • uuid (string: <required>) - Specifies the UUID of the query to explain. This is required and is specified as part of the URL path. This can also be the name of an existing prepared query, or a name that matches a prefix name for a prepared query template.

  • dc (string: "") - Specifies the datacenter to query. This will default to the datacenter of the agent being queried. This is specified as part of the URL as a query parameter.

» Sample Request 

$ curl \
    http://127.0.0.1:8500/v1/query/8f246b77-f3e1-ff88-5b48-8ec93abf3e05/explain

» Sample Response 

{
  "Query": {
    "ID": "8f246b77-f3e1-ff88-5b48-8ec93abf3e05",
    "Name": "my-query",
    "Session": "adf4238a-882b-9ddc-4a9d-5b6758e4159e",
    "Token": "<hidden>",
    "Name": "geo-db",
    "Template": {
      "Type": "name_prefix_match",
      "Regexp": "^geo-db-(.*?)-([^\\-]+?)$"
    },
    "Service": {
      "Service": "mysql-customer",
      "Failover": {
        "NearestN": 3,
        "Datacenters": ["dc1", "dc2"]
      },
      "OnlyPassing": true,
      "Tags": ["primary"],
      "Meta": { "mysql_version": "5.7.20" },
      "NodeMeta": {"instance_type": "m3.large"}
    }
  }
}