• API >
  • Public API Principles

Public API Principles


The Cloud Manager Public API follows the principles of the REST architectural style to expose a number of internal resources which enable programmatic access to Cloud Manager’s features.

As with changes made through the web interface, changes made through the API are subject to Cloud Manager pricing. If you add servers and incur charges, you must have a valid credit card on file with Cloud Manager or risk having your account locked.

The API has the following features:

  • JSON entities - All entities are expressed in JSON.
  • Digest authentication - To ensure that your Public API key is never sent over the network, API requests are authenticated using HTTP Digest Authentication.
  • Browsable interface - Using a consistent linking mechanism, you can browse the entire API by starting at the root resource and following links to related resources.
  • Security - In addition to Digest Authentication, the API is only accessible via HTTPS, and certain calls requiring even more security are protected by user-defined whitelists. Further, an API user’s capabilities are restricted by their assigned role(s). For example, a user with the Read Only role within a particular group will not be allowed to modify any resources within that group.

HTTP Methods

All resources support a subset of these common HTTP Methods:

  • GET - Retrieve the JSON representation of a resource.
  • POST - Create a new resource using the provided JSON representation.
  • PUT - Replace a resource with the provided JSON representation.
  • PATCH - Update the specified fields in a resource using the provided JSON representation.
  • DELETE - Remove a resource.
  • HEAD - Retrieve the response header without the JSON representation of the resource.


All entities are represented in JSON. The following rules and conventions apply:

  • When sending JSON to the server via POST or PUT, make sure to specify the correct content type request header: Content-Type: application/json
  • Invalid fields will be rejected rather than ignored. If, for example, you attempt to create a new entity and misspell one of the fields, or if you attempt to update an existing entity and include a field that cannot be modified, the server will respond with a 400 status code and an error message stating which field was invalid.
  • All dates are returned as ISO-8601 formatted strings designated in UTC. When sending dates to the server (ie, as query parameters or fields in POST or PATCH request entities), use ISO-8601 formatted dates. If you do not specify a time zone, UTC is assumed. However, it is highly recommended that you include a time zone designator to avoid any ambiguity.
  • In some cases, a timestamp will be returned as a BSON timestamp, most notably in the backup resources. These are represented in JSON documents as an object with two fields: date, an ISO-8601 formatted date string in UTC with granularity to the second, and increment a 32-bit integer.
  • Fields that contain numeric values in a particular unit will be named so as to disambiguate the unit being used. For example, a host’s uptime is returned in millseconds, so the name of the host entity field is uptimeMsec.
  • Fields that do not have a current value will be returned with an appropriate default value. For example, Cloud Manager will not have any statistics for a newly discovered host, so any statistics-related fields will have a value of zero. Fields that do not have a sensible default value will be omitted from the entity. For example, a host that is not using authentication will omit the username field from the returned entity.
  • The fields in the JSON documents returned by the server are in no particular order, and it may change. Do not depend on the order of the fields.


Each resource includes one or more links to sub-resources and/or related resources. For example, a host has a link to the group it belongs to, the replica set it belongs to, and so on. Links are placed in the links field of an entity, which is an array of link relation objects. Each link relation has two fields:

  • rel - Name (or type) of the relation. Many of these are considered Extension Relation Types and will be prefixed by
  • href - The target URL.

All entities include at least one link relation called self, which is simply its own URL. When an entity is part of a list (ie, when requesting all hosts in a group), then it only includes the self link relation. Here’s an example of a portion of a host resource with a few links:

  "id": "xxx",
  "groupId": "yyy",
  "hostname": "",
  "port": 27017,
  // additional host properties...
  "links": [
      "rel": "self",
      "href": ""
      "rel": "",
      "href": ""

For more information, refer to the Web Linking Specification. Note that although the specification describes a format for including links in the HTTP response headers, doing so is not a requirement. To make the API easily browsable, it includes the links in the response body rather than in the response headers.


Some resources return a list of entities. For example, you can request a list of all hosts in a group. When a list of entities is expected in a response, the results will be returned in batches bounded by two query parameters:

  • pageNum - Page number (1-based). Defaults to 1 if not specified.
  • itemsPerPage - Maximum number of items to return, up to a maximum of 100. Defaults to 100 if not specified.

The response entity contains three fields:

  • totalCount - The total number of items in the entire result set. For example, if a group has a total of 57 hosts, and you make a request with pageNum=6 and itemsPerPage=10, then totalCount will be 57.
  • results - The result set, which is an array of entity documents.
  • links - Contains one to three link relations: previous for the previous page of results (omitted for the first page); next for the next page of results (omitted for the last page); self for the current page (always present).

If you make a request for a list of entities and there are no results, then the API will respond with a 200 status code and the results array will be empty. It does not respond with a 404 in this case, since the list of entities may not be empty at some point in the future. However, had you requested a list of entities in a context that does not exist (i.e., the list of hosts for a non-existent group), then this will result in a 404 response status.

Here’s an example response for the second page of 10 hosts in a group with a total of 57 hosts:


  "totalCount": 57,
  "results": [
      "id": "yyy",
      "groupId": "xxx",
      // additional host properties...
    // additional host documents...
  "links": [
      "rel" : "self",
      "href" : ""
      "rel": "previous",
      "href": ""
      "rel": "next",
      "href": ""


Some clients may not be able to access the HTTP response headers and/or status code. In that case, you can request that the response include an “envelope,” which is simply an extra layer of information in the JSON document that contains any relevant details that would normally be in the response headers. By default, the API will not include the response in an envelope. To request one, simply add the query parameter envelope=true.

For responses that contain a single entity, the envelope will contain two fields:

  • status - The HTTP status code.
  • content - The requested entity.

For responses that contain a list of entities, there is already an envelope that wraps the results, so specifying envelope=true in this case will only add the status field to the existing envelope.

Pretty Printing

By default, extraneous whitespace is stripped from the JSON returned by the server. To ask for pretty-printed JSON, simply append the pretty=true query parameter to any request:

curl -u "username:apiKey" --digest -i ""


All the examples in this document show pretty-printed JSON for clarity, although the example URLs do not contain this additional query parameter.

Response Codes

Responses utilize the standard HTTP response codes, including:

Code Meaning Notes
200 OK The request was successful. This is typically the response to a successful GET request.
201 Created A new resource was created. This is typically the response to a successful POST request.
202 Accepted A request for an asynchronous operation was accepted.
400 Bad Request Something was wrong with the client request.
401 Unauthorized Authentication is required but was not present in the request. Typically this means that the digest authentication information was omitted from the request, the provided credentials are incorrect, or the user associated with the given API key is not allowed to access the requested resource.
403 Forbidden Access to the specified resource is not permitted.
404 Not Found The requested resource does not exist.
405 Method Not Allowed The HTTP method is not supported for the specified resource. Keep in mind that each resource may only support a subset of HTTP methods. For example, you are not allowed to DELETE the root resource.
409 Conflict This is typically the response to a request to create or modify a property of an entity that is unique when an existing entity already exists with the same value for that property. For example, attempting to create a group with the same name as an existing group is not allowed.
5xx Various server errors Something unexpected went wrong. Try again later and consider notifying Cloud Manager Support.


When a request results in an error, the response body will contain a document with additional details about what went wrong. The document contains three fields:

  • error - The error code, which is simply the HTTP status code.
  • reason - A short description of the error, which is simply the HTTP status phrase.
  • detail - A more detailed description of the error.

For example, here is the response body for a request for a host that does not exist:

  "error": 404,
  "reason": "Not Found",
  "detail": "No host exists with ID yyy in group xxx."

For a list of codes, see Public API Error Codes.


As previously mentioned, the Cloud Manager API uses HTTP Digest Authentication. The details of digest authentication are beyond the scope of this document, but it essentially requires a username and a password which are hashed using a unique server-generated value called a nonce. The username is the username of a registered Cloud Manager account, and the password is a Public API key associated with that account.

Keep the following points in mind:

  • The server-generated nonce is used by the client to hash the username and password before sending them back to the server to authenticate a request. The nonce is only valid for a short amount of time as per the digest authentication specification. This is to prevent replay attacks, so you can’t cache a nonce and use it forever.
  • Using digest authentication in combination with HTTPS provides an extra layer of security by ensuring that a password is never transmitted back to the server.
  • Some resource methods require even more security and are additionally protected by whitelists that allow access to the resource only from the IP addresses listed. Each user configures their own whitelist of IP addresses that allow access to the resource.
  • The Cloud Manager UI has a concept of roles, which allow more fine-grained control of the operations a user is allowed to perform. The API resources also enforce the same authorization rules, so the resources and methods that can be accessed by an API key are governed by the roles granted to the associated user. For example, to DELETE a host, the user that owns the API key used to make the request must be a Monitoring Admin or Owner in the group that the host belongs to.
  • Many resources are tied to a group, as evidenced by URLs of the form .../api/public/v1.0/groups/<GROUP-ID>/.... For these resources, the user tied to the API key must be a member of the group. Otherwise the server will respond with a 401 error.


The Automation Configuration Resource and Automation Status resources provide endpoints that let you modify a group’s deployment and retrieve deployment status. You can modify a deployment by sending a new automation configuration to Cloud Manager. The automation configuration is where you describe and configure the MongoDB processes to be deployed. Cloud Manager refers to this as the deployment’s “goal state.” When you submit a new automation configuration through the API, the Automation Agents adjust the current state of the system to match the goal state.


There is no protection in the API to prevent concurrent modifications. If two administrators both start with a configuration based on the current version, make their own modifications, and then submit their modifications, the later modification wins.

Rate Limiting

Certain resources are subject to rate limiting.

For resources that are rate limited, Cloud Manager will allow up to 100 requests per minute per group. Keep in mind that an API key is assigned to a user, but that user may access multiple groups. Consider two users: A and B. User A belongs to group X, and user B belongs to groups X and Y. At 1:00:00pm, User A makes 50 requests to a rate limited resource in group X, all of which are complete by 1:00:20pm. At 1:00:30pm, User B attempts to make 60 requests to a rate limited resource in group X. Since User A has already used up 50 requests within the 1:00pm minute for group X, the last 10 requests User B attempts to make will be rejected. However, User B can make requests to a rate limited resource in group Y, since each group maintains a separate request counter. At 1:01pm, requests to group X may proceed, because the request counter used for rate limiting will reset each minute.

If you exceed the rate limit, the API returns a 429 Too Many Requests response code.

Additional Information

See Public API Resources for a complete reference of all resources available in the Cloud Manager Public API.

←   API Public API Resources  →