open-consul/website/source/docs/agent/http.html.markdown

28 KiB

layout page_title sidebar_current
docs HTTP API docs-agent-http

HTTP API

The main interface to Consul is a RESTful HTTP API. The API can be used for CRUD for nodes, services, checks, and configuration. The endpoints are versioned to enable changes without breaking backwards compatibility.

All endpoints fall into one of 5 categories:

  • kv - Key/Value store
  • agent - Agent control
  • catalog - Manages nodes and services
  • health - Manages health checks
  • status - Consul system status

Each of the categories and their respective endpoints are documented below.

Blocking Queries

Certain endpoints support a feature called a "blocking query." A blocking query is used to wait for a change to potentially take place using long polling.

Queries that support this will mention it specifically, however the use of this feature is the same for all. If supported, the query will set an HTTP header "X-Consul-Index". This is an opaque handle that the client will use.

To cause a query to block, the query parameters "?wait=<interval>&index=<idx>" are added to a request. The "?wait=" query parameter limits how long the query will potentially block for. It not set, it will default to 10 minutes. It can be specified in the form of "10s" or "5m", which is 10 seconds or 5 minutes respectively. The "?index=" parameter is an opaque handle, which is used by Consul to detect changes. The "X-Consul-Index" header for a query provides this value, and can be used to wait for changes since the query was run.

When provided, Consul blocks sending a response until there is an update that could have cause the output to change, and thus advancing the index. A critical note is that when the query returns there is no guarantee of a change. It is possible that the timeout was reached, or that there was an idempotent write that does not affect the result.

Consistency Modes

Most of the read query endpoints support multiple levels of consistency. These are to provide a tuning knob that clients can be used to find a happy medium that best matches their needs.

The three read modes are:

  • default - If not specified, this mode is used. It is strongly consistent in almost all cases. However, there is a small window in which an new leader may be elected, and the old leader may service stale values. The trade off is fast reads, but potentially stale values. This condition is hard to trigger, and most clients should not need to worry about the stale read. This only applies to reads, and a split-brain is not possible on writes.

  • consistent - This mode is strongly consistent without caveats. It requires that a leader verify with a quorum of peers that it is still leader. This introduces an additional round-trip to all server nodes. The trade off is always consistent reads, but increased latency due to an extra round trip. Most clients should not use this unless they cannot tolerate a stale read.

  • stale - This mode allows any server to service the read, regardless of if it is the leader. This means reads can be arbitrarily stale, but are generally within 50 milliseconds of the leader. The trade off is very fast and scalable reads but values will be stale. This mode allows reads without a leader, meaning a cluster that is unavailable will still be able to respond.

To switch these modes, either the "?stale" or "?consistent" query parameters are provided. It is an error to provide both.

To support bounding how stale data is, there is an "X-Consul-LastContact" which is the last time a server was contacted by the leader node in milliseconds. The "X-Consul-KnownLeader" also indicates if there is a known leader. These can be used to gauge if a stale read should be used.

KV

The KV endpoint is used to expose a simple key/value store. This can be used to store service configurations or other meta data in a simple way. It has only a single endpoint:

/v1/kv/<key>

This is the only endpoint that is used with the Key/Value store. Its use depends on the HTTP method. The GET, PUT and DELETE methods are all supported. It is important to note that each datacenter has its own K/V store, and that there is no replication between datacenters. By default the datacenter of the agent is queried, however the dc can be provided using the "?dc=" query parameter. If a client wants to write to all Datacenters, one request per datacenter must be made.

When using the GET method, Consul will return the specified key, or if the "?recurse" query parameter is provided, it will return all keys with the given prefix.

Each object will look like:

[
    {
        "CreateIndex":100,
        "ModifyIndex":200,
        "Key":"zip",
        "Flags":0,
        "Value":"dGVzdA=="
    }
]

The CreateIndex is the internal index value that represents when the entry was created. The ModifyIndex is the last index that modified this key. This index corresponds to the X-Consul-Index header value that is returned. A blocking query can be used to wait for a value to change. If "?recurse" is used, the X-Consul-Index corresponds to the latest ModifyIndex and so a blocking query waits until any of the listed keys are updated. The multiple consistency modes can be used for GET requests as well.

The Key is simply the full path of the entry. Flags are an opaque unsigned integer that can be attached to each entry. The use of this is left totally to the user. Lastly, the Value is a base64 key value.

If no entries are found, a 404 code is returned.

When using the PUT method, Consul expects the request body to be the value corresponding to the key. There are a number of parameters that can be used with a PUT request:

  • ?flags=<num> : This can be used to specify an unsigned value between 0 and 2^64-1. It is opaque to the user, but a client application may use it.

  • ?cas=<index> : This flag is used to turn the PUT into a Check-And-Set operation. This is very useful as it allows clients to build more complex syncronization primitives on top. If the index is 0, then Consul will only put the key if it does not already exist. If the index is non-zero, then the key is only set if the index matches the ModifyIndex of that key.

The return value is simply either true or false. If the CAS check fails, then false will be returned.

Lastly, the DELETE method can be used to delete a single key or all keys sharing a prefix. If the "?recurse" query parameter is provided, then all keys with the prefix are deleted, otherwise only the specified key.

Agent

The Agent endpoints are used to interact with a local Consul agent. Usually, services and checks are registered with an agent, which then takes on the burden of registering with the Catalog and performing anti-entropy to recover from outages. There are also various control APIs that can be used instead of the msgpack RPC protocol.

The following endpoints are supported:

  • /v1/agent/checks: Returns the checks the local agent is managing
  • /v1/agent/services : Returns the services local agent is managing
  • /v1/agent/members : Returns the members as seen by the local serf agent
  • /v1/agent/join/<address> : Trigger local agent to join a node
  • /v1/agent/force-leave/<node>: Force remove node
  • /v1/agent/check/register : Registers a new local check
  • /v1/agent/check/deregister/<checkID> : Deregister a local check
  • /v1/agent/check/pass/<checkID> : Mark a local test as passing
  • /v1/agent/check/warn/<checkID> : Mark a local test as warning
  • /v1/agent/check/fail/<checkID> : Mark a local test as critical
  • /v1/agent/service/register : Registers a new local service
  • /v1/agent/service/deregister/<serviceID> : Deregister a local service

/v1/agent/checks

This endpoint is used to return the all the checks that are registered with the local agent. These checks were either provided through configuration files, or added dynamically using the HTTP API. It is important to note that the checks known by the agent may be different than those reported by the Catalog. This is usually due to changes being made while there is no leader elected. The agent performs active anti-entropy, so in most situations everything will be in sync within a few seconds.

This endpoint is hit with a GET and returns a JSON body like this:

{
    "service:redis":{
        "Node":"foobar",
        "CheckID":"service:redis",
        "Name":"Service 'redis' check",
        "Status":"passing",
        "Notes":"",
        "ServiceID":"redis",
        "ServiceName":"redis"
    }
}

/v1/agent/services

This endpoint is used to return the all the services that are registered with the local agent. These services were either provided through configuration files, or added dynamically using the HTTP API. It is important to note that the services known by the agent may be different than those reported by the Catalog. This is usually due to changes being made while there is no leader elected. The agent performs active anti-entropy, so in most situations everything will be in sync within a few seconds.

This endpoint is hit with a GET and returns a JSON body like this:

{
    "redis":{
        "ID":"redis",
        "Service":"redis",
        "Tags":null,
        "Port":8000
    }
}

/v1/agent/members

This endpoint is hit with a GET and returns the members the agent sees in the cluster gossip pool. Due to the nature of gossip, this is eventually consistent and the results may differ by agent. The strongly consistent view of nodes is instead provided by "/v1/catalog/nodes".

For agents running in server mode, providing a "?wan=1" query parameter returns the list of WAN members instead of the LAN members which is default.

This endpoint returns a JSON body like:

[
    {
        "Name":"foobar",
        "Addr":"10.1.10.12",
        "Port":8301,
        "Tags":{
            "bootstrap":"1",
            "dc":"dc1",
            "port":"8300",
            "role":"consul"
        },
        "Status":1,
        "ProtocolMin":1,
        "ProtocolMax":2,
        "ProtocolCur":2,
        "DelegateMin":1,
        "DelegateMax":3,
        "DelegateCur":3
    }
]

/v1/agent/join/<address>

This endpoint is hit with a GET and is used to instruct the agent to attempt to connect to a given address. For agents running in server mode, providing a "?wan=1" query parameter causes the agent to attempt to join using the WAN pool.

The endpoint returns 200 on successful join.

/v1/agent/force-leave/<node>

This endpoint is hit with a GET and is used to instructs the agent to force a node into the left state. If a node fails unexpectedly, then it will be in a "failed" state. Once in this state, Consul will attempt to reconnect, and additionally the services and checks belonging to that node will not be cleaned up. Forcing a node into the left state allows its old entries to be removed.

The endpoint always returns 200.

/v1/agent/check/register

The register endpoint is used to add a new check to the local agent. There is more documentation on checks here. Checks are either a script or TTL type. The agent is reponsible for managing the status of the check and keeping the Catalog in sync.

The register endpoint expects a JSON request body to be PUT. The request body must look like:

{
    "ID": "mem",
    "Name": "Memory utilization",
    "Notes": "Ensure we don't oversubscribe memory",
    "Script": "/usr/local/bin/check_mem.py",
    "Interval": "10s",
    "TTL": "15s"
}

The Name field is mandatory, as is either Script and Interval or TTL. Only one of Script and Interval or TTL should be provided. If an ID is not provided, it is set to Name. You cannot have duplicate ID entries per agent, so it may be necessary to provide an ID. The Notes field is not used by Consul, and is meant to be human readable.

If a Script is provided, the check type is a script, and Consul will evaluate the script every Interval to update the status. If a TTL type is used, then the TTL update APIs must be used to periodically update the state of the check.

The return code is 200 on success.

/v1/agent/check/deregister/<checkId>

The deregister endpoint is used to remove a check from the local agent. The CheckID must be passed after the slash. The agent will take care of deregistering the check with the Catalog.

The return code is 200 on success.

/v1/agent/check/pass/<checkId>

This endpoint is used with a check that is of the TTL type. When this endpoint is accessed, the status of the check is set to "passing", and the TTL clock is reset.

The return code is 200 on success.

/v1/agent/check/warn/<checkId>

This endpoint is used with a check that is of the TTL type. When this endpoint is accessed, the status of the check is set to "warning", and the TTL clock is reset.

The return code is 200 on success.

/v1/agent/check/fail/<checkId>

This endpoint is used with a check that is of the TTL type. When this endpoint is accessed, the status of the check is set to "critical", and the TTL clock is reset.

The return code is 200 on success.

/v1/agent/service/register

The register endpoint is used to add a new service to the local agent. There is more documentation on services here. Services may also provide a health check. The agent is reponsible for managing the status of the check and keeping the Catalog in sync.

The register endpoint expects a JSON request body to be PUT. The request body must look like:

{
    "ID": "redis1",
    "Name": "redis",
    "Tags": ["master", "v1"],
    "Port": 8000,
    "Check": {
        "Script": "/usr/local/bin/check_redis.py",
        "Interval": "10s",
        "TTL": "15s"
    }
}

The Name field is mandatory, If an ID is not provided, it is set to Name. You cannot have duplicate ID entries per agent, so it may be necessary to provide an ID. Tags, Port and Check are optional. If Check is provided, only one of Script and Interval or TTL should be provided. There is more information about checks here.

The created check will be named "service:<ServiceId>".

The return code is 200 on success.

/v1/agent/service/deregister/<serviceId>

The deregister endpoint is used to remove a service from the local agent. The ServiceID must be passed after the slash. The agent will take care of deregistering the service with the Catalog. If there is an associated check, that is also deregistered.

The return code is 200 on success.

Catalog

The Catalog is the endpoint used to register and deregister nodes, services, and checks. It also provides a number of query endpoints.

The following endpoints are supported:

  • /v1/catalog/register : Registers a new node, service, or check
  • /v1/catalog/deregister : Deregisters a node, service, or check
  • /v1/catalog/datacenters : Lists known datacenters
  • /v1/catalog/nodes : Lists nodes in a given DC
  • /v1/catalog/services : Lists services in a given DC
  • /v1/catalog/service/<service> : Lists the nodes in a given service
  • /v1/catalog/node/<node> : Lists the services provided by a node

The last 4 endpoints of the catalog support blocking queries and consistency modes.

/v1/catalog/register

The register endpoint is a low level mechanism for direclty registering or updating entries in the catalog. It is usually recommended to use the agent local endpoints, as they are simpler and perform anti-entropy.

The register endpoint expects a JSON request body to be PUT. The request body must look like:

{
    "Datacenter": "dc1",
    "Node": "foobar",
    "Address": "192.168.10.10",
    "Service": {
        "ID": "redis1",
        "Service": "redis",
        "Tags": ["master", "v1"],
        "Port": 8000,
    },
    "Check": {
        "Node": "foobar",
        "CheckID": "service:redis1",
        "Name": "Redis health check",
        "Notes": "Script based health check",
        "Status": "passing",
        "ServiceID": "redis1"
    }
}

The behavior of the endpoint depends on what keys are provided. The endpoint requires Node and Address to be provided, while Datacenter will be defaulted to match that of the agent. If only those are provided, the endpoint will register the node with the catalog.

If the Service key is provided, then the service will also be registered. If ID is not provided, it will be defaulted to Service. It is mandated that the ID be node-unique. Both Tags and Port can be omitted.

If the Check key is provided, then a health check will also be registered. It is important to remember that this register API is very low level. This manipulates the health check entry, but does not setup a script or TTL to actually update the status. For that behavior, an agent local check should be setup.

The CheckID can be omitted, and will default to the Name. Like before, the CheckID must be node-unique. The Notes is an opaque field that is meant to hold human readable text. If a ServiceID is provided that matches the ID of a service on that node, then the check is treated as a service level health check, instead of a node level health check. Lastly, the status must be one of "unknown", "passing", "warning", or "critical". The "unknown" status is used to indicate that the initial check has not been performed yet.

It is important to note that Check does not have to be provided with Service and visa-versa. They can be provided or omitted at will.

If the API call succeeds a 200 status code is returned.

/v1/catalog/deregister

The deregister endpoint is a low level mechanism for direclty removing entries in the catalog. It is usually recommended to use the agent local endpoints, as they are simpler and perform anti-entropy.

The deregister endpoint expects a JSON request body to be PUT. The request body must look like one of the following:

{
    "Datacenter": "dc1",
    "Node": "foobar",
}

{
    "Datacenter": "dc1",
    "Node": "foobar",
    "CheckID": "service:redis1"
}

{
    "Datacenter": "dc1",
    "Node": "foobar",
    "ServiceID": "redis1",
}

The behavior of the endpoint depends on what keys are provided. The endpoint requires Node to be provided, while Datacenter will be defaulted to match that of the agent. If only Node is provided, then the node, and all associated services and checks are deleted. If CheckID is provided, only that check belonging to the node is removed. If ServiceID is provided, then the service along with it's associated health check (if any) is removed.

If the API call succeeds a 200 status code is returned.

/v1/catalog/datacenters

This endpoint is hit with a GET and is used to return all the datacenters that are known by the Consul server.

It returns a JSON body like this:

["dc1", "dc2"]

This endpoint does not require a cluster leader, and as such will succeed even during an availability outage. It can thus be a simple check to see if any Consul servers are routable.

/v1/catalog/nodes

This endpoint is hit with a GET and returns the nodes known about in a given DC. By default the datacenter of the agent is queried, however the dc can be provided using the "?dc=" query parameter.

It returns a JSON body like this:

[
    {
        "Node":"baz",
        "Address":"10.1.10.11"
    },
    {
        "Node":"foobar",
        "Address":"10.1.10.12"
    }
]

This endpoint supports blocking queries and all consistency modes.

/v1/catalog/services

This endpoint is hit with a GET and returns the services known about in a given DC. By default the datacenter of the agent is queried, however the dc can be provided using the "?dc=" query parameter.

It returns a JSON body like this:

{
    "consul":[],
    "redis":[],
    "postgresql":["master","slave"]
}

The main object keys are the service names, while the array provides all the known tags for a given service.

This endpoint supports blocking queries and all consistency modes.

/v1/catalog/service/<service>

This endpoint is hit with a GET and returns the nodes providing a service in a given DC. By default the datacenter of the agent is queried, however the dc can be provided using the "?dc=" query parameter.

The service being queried must be provided after the slash. By default all nodes in that service are returned. However, the list can be filtered by tag using the "?tag=" query parameter.

It returns a JSON body like this:

[
    {
        "Node":"foobar",
        "Address":"10.1.10.12",
        "ServiceID":"redis",
        "ServiceName":"redis",
        "ServiceTags":null,
        "ServicePort":8000
    }
]

This endpoint supports blocking queries and all consistency modes.

/v1/catalog/node/<node>

This endpoint is hit with a GET and returns the node provided services. By default the datacenter of the agent is queried, however the dc can be provided using the "?dc=" query parameter. The node being queried must be provided after the slash.

It returns a JSON body like this:

{
    "Node":{
        "Node":"foobar",
        "Address":"10.1.10.12"
    },
    "Services":{
        "consul":{
            "ID":"consul",
            "Service":"consul",
            "Tags":null,
            "Port":8300
        },
        "redis":{
            "ID":"redis",
            "Service":"redis",
            "Tags":["v1"],
            "Port":8000
        }
    }
}

This endpoint supports blocking queries and all consistency modes.

Health

The Health used to query health related information. It is provided seperately from the Catalog, since users may prefer to not use the health checking mechanisms as they are totally optional. Additionally, some of the query results from the Health system are filtered, while the Catalog endpoints provide the raw entries.

The following endpoints are supported:

  • /v1/health/node/<node>: Returns the health info of a node
  • /v1/health/checks/<service>: Returns the checks of a service
  • /v1/health/service/<service>: Returns the nodes and health info of a service
  • /v1/health/state/<state>: Returns the checks in a given state

All of the health endpoints supports blocking queries and all consistency modes.

/v1/health/node/<node>

This endpoint is hit with a GET and returns the node specific checks known. By default the datacenter of the agent is queried, however the dc can be provided using the "?dc=" query parameter. The node being queried must be provided after the slash.

It returns a JSON body like this:

[
    {
        "Node":"foobar",
        "CheckID":"serfHealth",
        "Name":"Serf Health Status",
        "Status":"passing",
        "Notes":"",
        "ServiceID":"",
        "ServiceName":""
    },
    {
        "Node":"foobar",
        "CheckID":"service:redis",
        "Name":"Service 'redis' check",
        "Status":"passing",
        "Notes":"",
        "ServiceID":"redis",
        "ServiceName":"redis"
    }
]

In this case, we can see there is a system level check (no associated ServiceID, as well as a service check for Redis). The "serfHealth" check is special, in that all nodes automatically have this check. When a node joins the Consul cluster, it is part of a distributed failure detection provided by Serf. If a node fails, it is detected and the status is automatically changed to "critical".

This endpoint supports blocking queries and all consistency modes.

/v1/health/checks/<service>

This endpoint is hit with a GET and returns the checks associated with a service in a given datacenter. By default the datacenter of the agent is queried, however the dc can be provided using the "?dc=" query parameter. The service being queried must be provided after the slash.

It returns a JSON body like this:

[
    {
        "Node":"foobar",
        "CheckID":"service:redis",
        "Name":"Service 'redis' check",
        "Status":"passing",
        "Notes":"",
        "ServiceID":"redis",
        "ServiceName":"redis"
    }
]

This endpoint supports blocking queries and all consistency modes.

/v1/health/service/<service>

This endpoint is hit with a GET and returns the service nodes providing a given service in a given datacenter. By default the datacenter of the agent is queried, however the dc can be provided using the "?dc=" query parameter.

The service being queried must be provided after the slash. By default all nodes in that service are returned. However, the list can be filtered by tag using the "?tag=" query parameter.

This is very similar to the /v1/catalog/service endpoint however, this endpoint automatically returns the status of the associated health check, as well as any system level health checks. This allows a client to avoid sending traffic to nodes failing health tests, or who are reporting warnings.

Providing the "?passing" query parameter will filter results to only nodes with all checks in the passing state. This can be used to avoid some filtering logic on the client side. (Added in Consul 0.2)

Users can also built in support for dynamic load balancing and other features by incorporating the use of health checks.

It returns a JSON body like this:

[
    {
        "Node":{
            "Node":"foobar",
            "Address":"10.1.10.12"
        },
        "Service":{
            "ID":"redis",
            "Service":"redis",
            "Tags":null,
            "Port":8000
        },
        "Checks":[
            {
                "Node":"foobar",
                "CheckID":"service:redis",
                "Name":"Service 'redis' check",
                "Status":"passing",
                "Notes":"",
                "ServiceID":"redis",
                "ServiceName":"redis"
            },{
                "Node":"foobar",
                "CheckID":"serfHealth",
                "Name":"Serf Health Status",
                "Status":"passing",
                "Notes":"",
                "ServiceID":"",
                "ServiceName":""
            }
        ]
    }
]

This endpoint supports blocking queries and all consistency modes.

/v1/health/state/<state>

This endpoint is hit with a GET and returns the checks in a specific state for a given datacenter. By default the datacenter of the agent is queried, however the dc can be provided using the "?dc=" query parameter.

The state being queried must be provided after the slash. The supported states are "unknown", "passing", "warning", or "critical".

It returns a JSON body like this:

[
    {
        "Node":"foobar",
        "CheckID":"serfHealth",
        "Name":"Serf Health Status",
        "Status":"passing",
        "Notes":"",
        "ServiceID":"",
        "ServiceName":""
    },
    {
        "Node":"foobar",
        "CheckID":"service:redis",
        "Name":"Service 'redis' check",
        "Status":"passing",
        "Notes":"",
        "ServiceID":"redis",
        "ServiceName":"redis"
    }
]

This endpoint supports blocking queries and all consistency modes.

Status

The Status endpoints are used to get information about the status of the Consul cluster. This are generally very low level, and not really useful for clients.

The following endpoints are supported:

  • /v1/status/leader : Returns the current Raft leader
  • /v1/status/peers : Returns the current Raft peer set

/v1/status/leader

This endpoint is used to get the Raft leader for the datacenter the agent is running in. It returns only an address like:

"10.1.10.12:8300"

/v1/status/peers

This endpoint is used to get the Raft peers for the datacenter the agent is running in. It returns a list of addresses like:

["10.1.10.12:8300", "10.1.10.11:8300", "10.1.10.10:8300"]