--- layout: api page_title: HTTP API description: |- Nomad exposes a RESTful HTTP API to control almost every aspect of the Nomad agent. --- # HTTP API The main interface to Nomad is a RESTful HTTP API. The API can query the current state of the system as well as modify the state of the system. The Nomad CLI actually invokes Nomad's HTTP for many commands. ## Version Prefix All API routes are prefixed with `/v1/`. ## Addressing and Ports Nomad binds to a specific set of addresses and ports. The HTTP API is served via the `http` address and port. This `address:port` must be accessible locally. If you bind to `127.0.0.1:4646`, the API is only available _from that host_. If you bind to a private internal IP, the API will be available from within that network. If you bind to a public IP, the API will be available from the public Internet (not recommended). The default port for the Nomad HTTP API is `4646`. This can be overridden via the Nomad configuration block. Here is an example curl request to query a Nomad server with the default configuration: ```shell-session $ curl http://127.0.0.1:4646/v1/agent/members ``` The conventions used in the API documentation do not list a port and use the standard URL `localhost:4646`. Be sure to replace this with your Nomad agent URL when using the examples. ## Data Model and Layout There are five primary nouns in Nomad: - jobs - nodes - allocations - deployments - evaluations [![Nomad Data Model](/img/nomad-data-model.png)](/img/nomad-data-model.png) Jobs are submitted by users and represent a _desired state_. A job is a declarative description of tasks to run which are bounded by constraints and require resources. Jobs can also have affinities which are used to express placement preferences. Nodes are the servers in the clusters that tasks can be scheduled on. The mapping of tasks in a job to nodes is done using allocations. An allocation is used to declare that a set of tasks in a job should be run on a particular node. Scheduling is the process of determining the appropriate allocations and is done as part of an evaluation. Deployments are objects to track a rolling update of allocations between two versions of a job. The API is modeled closely on the underlying data model. Use the links to the left for documentation about specific endpoints. There are also "Agent" APIs which interact with a specific agent and not the broader cluster used for administration. ## ACLs Several endpoints in Nomad use or require ACL tokens to operate. The token are used to authenticate the request and determine if the request is allowed based on the associated authorizations. Tokens are specified per-request by using the `X-Nomad-Token` request header or with the Bearer scheme in the authorization header set to the `SecretID` of an ACL Token. For more details about ACLs, please see the [ACL Guide](https://learn.hashicorp.com/collections/nomad/access-control). ## Authentication When ACLs are enabled, a Nomad token should be provided to API requests using the `X-Nomad-Token` header or with the Bearer scheme in the authorization header. When using authentication, clients should communicate via TLS. Here is an example using curl with `X-Nomad-Token`: ```shell-session $ curl \ --header "X-Nomad-Token: aa534e09-6a07-0a45-2295-a7f77063d429" \ https://localhost:4646/v1/jobs ``` Below is an example using `curl` with a [RFC6750](https://tools.ietf.org/html/rfc6750) Bearer token: ```shell-session $ curl \ --header "Authorization: Bearer " \ http://localhost:4646/v1/jobs ``` ## Namespaces Nomad has support for namespaces, which allow jobs and their associated objects to be segmented from each other and other users of the cluster. When using non-default namespace, the API request must pass the target namespace as an API query parameter. Prior to Nomad 1.0 namespaces were Enterprise-only. Here is an example using curl: ```shell-session $ curl https://localhost:4646/v1/jobs?namespace=qa ``` ## Filtering Filter expressions refine data queries for some API listing endpoints, as notated in the individual API endpoints documentation. To create a filter expression, you will write one or more expressions. Each expression has matching operators composed of selectors and values. Filtering is executed on the Nomad server, before data is returned, reducing the network load. To pass a filter expression to Nomad, use the `filter` query parameter with the URL encoded expression when sending requests to HTTP API endpoints that support it. ```shell-session $ curl --get https://localhost:4646/v1/ --data-urlencode 'filter=' ``` Some endpoints may have other query parameters that are used for filtering, but they can't be used with the `filter` query parameter. Doing so will result in a `400` status error response. These query parameters are usually backed by a database index, so they may be prefereable over an equivalent simple `filter` expression due to better resource usage and performance. ### Creating Expressions A single expression is a matching operator with a selector and value and they are written in plain text format. Boolean logic and parenthesization are supported. In general, whitespace is ignored, except within literal strings. #### Matching Operators All matching operators use a selector or value to choose what data should be matched. Each endpoint that supports filtering accepts a potentially different list of selectors and is detailed in the API documentation for those endpoints. ```hcl // Equality & Inequality checks == "" != "" // Emptiness checks is empty is not empty // Contains checks or Substring Matching "" in "" not in contains "" not contains "" // Regular Expression Matching matches "" not matches "" ``` #### Selectors Selectors are used by matching operators to create an expression. They are defined by a `.` separated list of names. Each name must start with an ASCII letter and can contain ASCII letters, numbers, and underscores. When part of the selector references a map value it may be expressed using the form `[""]` instead of `.`. This allows the possibility of using map keys that are not valid selectors in and of themselves. ```hcl // selects the `cache` key within the `TaskGroups` mapping for the // /v1/deployments endpoint TaskGroups.cache // Also selects the `cache` key for the same endpoint TaskGroups["cache"] ``` #### Values Values are used by matching operators to create an expression. Values can be any valid selector, a number, or a string. It is best practice to quote values. Numbers can be base 10 integers or floating point numbers. When quoting strings, they may either be enclosed in double quotes or backticks. When enclosed in backticks they are treated as raw strings and escape sequences such as `\n` will not be expanded. ### Connecting Expressions There are several methods for connecting expressions, including: - logical `or` - logical `and` - logical `not` - grouping with parenthesis - matching expressions ```hcl // Logical Or - evaluates to true if either sub-expression does or // Logical And - evaluates to true if both sub-expressions do and // Logical Not - evaluates to true if the sub-expression does not not // Grouping - Overrides normal precedence rules ( ) // Inspects data to check for a match ``` Standard operator precedence can be expected for the various forms. For example, the following two expressions would be equivalent. ```hcl and not or ( and (not )) or ``` ### Filter Utilization Generally, only the main object is filtered. When filtering for an item within an array that is not at the top level, the entire array that contains the item will be returned. This is usually the outermost object of a response, but in some cases the filtering is performed on a object embedded within the results. #### Performance Filters are executed on the servers and therefore will consume some amount of CPU time on the server. For non-stale queries this means that the filter is executed on the leader. #### Filtering Examples ##### Jobs API Command (Unfiltered) ```shell-session $ curl --request GET https://localhost:4646/v1/jobs ``` Response (Unfiltered) ```json [ { "CreateIndex": 52, "Datacenters": [ "dc1", "dc2" ], "ID": "countdash", "JobModifyIndex": 56, "JobSummary": { "Children": { "Dead": 0, "Pending": 0, "Running": 0 }, "CreateIndex": 52, "JobID": "countdash", "ModifyIndex": 55, "Namespace": "default", "Summary": { "api": { "Complete": 0, "Failed": 0, "Lost": 0, "Queued": 1, "Running": 0, "Starting": 0 }, "dashboard": { "Complete": 0, "Failed": 0, "Lost": 0, "Queued": 1, "Running": 0, "Starting": 0 } } }, "ModifyIndex": 56, "Multiregion": null, "Name": "countdash", "Namespace": "default", "ParameterizedJob": false, "ParentID": "", "Periodic": false, "Priority": 50, "Status": "pending", "StatusDescription": "", "Stop": false, "SubmitTime": 1645230445788556000, "Type": "service" }, { "CreateIndex": 42, "Datacenters": [ "dc1" ], "ID": "example", "JobModifyIndex": 42, "JobSummary": { "Children": { "Dead": 0, "Pending": 0, "Running": 0 }, "CreateIndex": 42, "JobID": "example", "ModifyIndex": 46, "Namespace": "default", "Summary": { "cache": { "Complete": 0, "Failed": 0, "Lost": 0, "Queued": 0, "Running": 1, "Starting": 0 } } }, "ModifyIndex": 49, "Multiregion": null, "Name": "example", "Namespace": "default", "ParameterizedJob": false, "ParentID": "", "Periodic": false, "Priority": 50, "Status": "running", "StatusDescription": "", "Stop": false, "SubmitTime": 1645230403921889000, "Type": "service" } ] ``` Command (Filtered) ```shell curl --get https://localhost:4646/v1/jobs \ --data-urlencode 'filter=Datacenters contains "dc2"' ``` Response (Filtered) ```json [ { "CreateIndex": 52, "Datacenters": [ "dc1", "dc2" ], "ID": "countdash", "JobModifyIndex": 56, "JobSummary": { "Children": { "Dead": 0, "Pending": 0, "Running": 0 }, "CreateIndex": 52, "JobID": "countdash", "ModifyIndex": 55, "Namespace": "default", "Summary": { "api": { "Complete": 0, "Failed": 0, "Lost": 0, "Queued": 1, "Running": 0, "Starting": 0 }, "dashboard": { "Complete": 0, "Failed": 0, "Lost": 0, "Queued": 1, "Running": 0, "Starting": 0 } } }, "ModifyIndex": 56, "Multiregion": null, "Name": "countdash", "Namespace": "default", "ParameterizedJob": false, "ParentID": "", "Periodic": false, "Priority": 50, "Status": "pending", "StatusDescription": "", "Stop": false, "SubmitTime": 1645230445788556000, "Type": "service" } ] ``` ##### Deployments API Command (Unfiltered) ```shell-session $ curl --request GET https://localhost:4646/v1/deployments ``` Response (Unfiltered) ```json [ { "CreateIndex": 54, "EvalPriority": 50, "ID": "58fd0616-ce64-d14b-6917-03d0ab5af67e", "IsMultiregion": false, "JobCreateIndex": 52, "JobID": "countdash", "JobModifyIndex": 52, "JobSpecModifyIndex": 52, "JobVersion": 0, "ModifyIndex": 59, "Namespace": "default", "Status": "cancelled", "StatusDescription": "Cancelled due to newer version of job", "TaskGroups": { "dashboard": { "AutoPromote": false, "AutoRevert": false, "DesiredCanaries": 0, "DesiredTotal": 1, "HealthyAllocs": 0, "PlacedAllocs": 0, "PlacedCanaries": null, "ProgressDeadline": 600000000000, "Promoted": false, "RequireProgressBy": null, "UnhealthyAllocs": 0 }, "api": { "AutoPromote": false, "AutoRevert": false, "DesiredCanaries": 0, "DesiredTotal": 1, "HealthyAllocs": 0, "PlacedAllocs": 0, "PlacedCanaries": null, "ProgressDeadline": 600000000000, "Promoted": false, "RequireProgressBy": null, "UnhealthyAllocs": 0 } } }, { "CreateIndex": 43, "EvalPriority": 50, "ID": "1f18b48c-b33b-8e96-5640-71e3f3000242", "IsMultiregion": false, "JobCreateIndex": 42, "JobID": "example", "JobModifyIndex": 42, "JobSpecModifyIndex": 42, "JobVersion": 0, "ModifyIndex": 49, "Namespace": "default", "Status": "successful", "StatusDescription": "Deployment completed successfully", "TaskGroups": { "cache": { "AutoPromote": false, "AutoRevert": false, "DesiredCanaries": 0, "DesiredTotal": 1, "HealthyAllocs": 1, "PlacedAllocs": 1, "PlacedCanaries": null, "ProgressDeadline": 600000000000, "Promoted": false, "RequireProgressBy": "2022-02-18T19:36:54.421823-05:00", "UnhealthyAllocs": 0 } } } ] ``` Command (Filtered) ```shell curl --get https://localhost:4646/v1/deployments \ --data-urlencode 'filter=Status != "successful"' ``` Response (Filtered) ```json [ { "CreateIndex": 54, "EvalPriority": 50, "ID": "58fd0616-ce64-d14b-6917-03d0ab5af67e", "IsMultiregion": false, "JobCreateIndex": 52, "JobID": "countdash", "JobModifyIndex": 52, "JobSpecModifyIndex": 52, "JobVersion": 0, "ModifyIndex": 59, "Namespace": "default", "Status": "cancelled", "StatusDescription": "Cancelled due to newer version of job", "TaskGroups": { "dashboard": { "AutoPromote": false, "AutoRevert": false, "DesiredCanaries": 0, "DesiredTotal": 1, "HealthyAllocs": 0, "PlacedAllocs": 0, "PlacedCanaries": null, "ProgressDeadline": 600000000000, "Promoted": false, "RequireProgressBy": null, "UnhealthyAllocs": 0 }, "api": { "AutoPromote": false, "AutoRevert": false, "DesiredCanaries": 0, "DesiredTotal": 1, "HealthyAllocs": 0, "PlacedAllocs": 0, "PlacedCanaries": null, "ProgressDeadline": 600000000000, "Promoted": false, "RequireProgressBy": null, "UnhealthyAllocs": 0 } } } ] ``` ## Pagination Some list endpoints support partial results to limit the amount of data retrieved. The returned list is split into pages and the page size can be set using the `per_page` query parameter with a positive integer value. If more data is available past the page requested, the response will contain an HTTP header named `X-Nomad-Nexttoken` with the value of the next item to be retrieved. This value can then be set as a query parameter called `next_token` in a follow-up request to retrieve the next page. When the last page is reached, the `X-Nomad-Nexttoken` HTTP header will not be present in the response, indicating that there is nothing more to return. ## Ordering List results are usually returned in ascending order by their internal key, such as their `ID`. Some endpoints may return data sorted by their `CreateIndex` value, which roughly corelates to their creation order. The result order may be reversed using the `reverse=true` query parameter when supported by the endpoint. ## Blocking Queries Many endpoints in Nomad support a feature known as "blocking queries". A blocking query is used to wait for a potential change using long polling. Not all endpoints support blocking, but each endpoint uniquely documents its support for blocking queries in the documentation. Endpoints that support blocking queries return an HTTP header named `X-Nomad-Index`. This is a unique identifier representing the current state of the requested resource. On a new Nomad cluster the value of this index starts at 1. On subsequent requests for this resource, the client can set the `index` query string parameter to the value of `X-Nomad-Index`, indicating that the client wishes to wait for any changes subsequent to that index. When this is provided, the HTTP request will "hang" until a change in the system occurs, or the maximum timeout is reached. A critical note is that the return of a blocking request 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 of the query. In addition to `index`, endpoints that support blocking will also honor a `wait` parameter specifying a maximum duration for the blocking request. This is limited to 10 minutes. If not set, the wait time defaults to 5 minutes. This value can be specified in the form of "10s" or "5m" (i.e., 10 seconds or 5 minutes, respectively). A small random amount of additional wait time is added to the supplied maximum `wait` time to spread out the wake up time of any concurrent requests. This adds up to `wait / 16` additional time to the maximum duration. ## Consistency Modes Most of the read query endpoints support multiple levels of consistency. Since no policy will suit all clients' needs, these consistency modes allow the user to have the ultimate say in how to balance the trade-offs inherent in a distributed system. The two read modes are: - `default` - If not specified, the default is strongly consistent in almost all cases. However, there is a small window in which a new leader may be elected during which the old leader may service stale values. The trade-off is fast reads but potentially stale values. The condition resulting in stale reads is hard to trigger, and most clients should not need to worry about this case. Also, note that this race condition only applies to reads, not writes. - `stale` - This mode allows any server to service the read regardless of whether it is the leader. This means reads can be arbitrarily stale; however, results are generally consistent to within 50 milliseconds of the leader. The trade-off is very fast and scalable reads with a higher likelihood of stale values. Since this mode allows reads without a leader, a cluster that is unavailable will still be able to respond to queries. To switch these modes, use the `stale` query parameter on requests. To support bounding the acceptable staleness of data, responses provide the `X-Nomad-LastContact` header containing the time in milliseconds that a server was last contacted by the leader node. The `X-Nomad-KnownLeader` header also indicates if there is a known leader. These can be used by clients to gauge the staleness of a result and take appropriate action. ## Cross-Region Requests By default, any request to the HTTP API will default to the region on which the machine is servicing the request. If the agent runs in "region1", the request will query the region "region1". A target region can be explicitly request using the `?region` query parameter. The request will be transparently forwarded and serviced by a server in the requested region. ## Compressed Responses The HTTP API will gzip the response if the HTTP request denotes that the client accepts gzip compression. This is achieved by passing the accept encoding: ```shell-session $ curl \ --header "Accept-Encoding: gzip" \ https://localhost:4646/v1/... ``` ## Formatted JSON Output By default, the output of all HTTP API requests is minimized JSON. If the client passes `pretty` on the query string, formatted JSON will be returned. In general, clients should prefer a client-side parser like `jq` instead of server-formatted data. Asking the server to format the data takes away processing cycles from more important tasks. ```shell-session $ curl https://localhost:4646/v1/page?pretty ``` ## HTTP Methods Nomad's API aims to be RESTful, although there are some exceptions. The API responds to the standard HTTP verbs GET, PUT, and DELETE. Each API method will clearly document the verb(s) it responds to and the generated response. The same path with different verbs may trigger different behavior. For example: ```text PUT /v1/jobs GET /v1/jobs ``` Even though these share a path, the `PUT` operation creates a new job whereas the `GET` operation reads all jobs. ## HTTP Response Codes Individual API's will contain further documentation in the case that more specific response codes are returned but all clients should handle the following: - 200 and 204 as success codes. - 400 indicates a validation failure and if a parameter is modified in the request, it could potentially succeed. - 403 marks that the client isn't authenticated for the request. - 404 indicates an unknown resource. - 5xx means that the client should not expect the request to succeed if retried.