Interface

Workbench API

Method launch_session

Launches a Posit Workbench Session on behalf of the username provided in the parameters or the auth cookie. A super-admin API token is required to launch a session on behalf of a different user than the API token owner.

POST /api/launch_session {
   "method": "launch_session",
   "kwparams": {
      "workbench": "RStudio",
      "name": "My RStudio Session",
      "launch_parameters": {
         "name": "Job name", "cluster": "Local",
         "placement_constraints":[], "resource_limits":[], "queues":[]
      }
   }
}

Parameters

Name	Type	Summary
workbench (required)	string	The IDE name to use: one of: `RStudio`, `JupyterLab`, `Jupyter Notebook`, `VS Code`.
username	string	Start the session on behalf of this user. If not provided, the auth-token’s username is used.
name	string	The new session’s name, as it appears in the homepage.
open_file	string	A file to open by default in the launched IDE session. This is supported for RStudio and Jupyter Notebook sessions.
working_directory	string	The current directory for the new session.
launch_parameters	LaunchParameters	Provides additional launch config, required when launcher sessions are enabled.
start_hooks	SessionHook	Provides hooks that should be run when the session is started.
stop_hooks	SessionHook	Provides hooks that should be run when the session is stopped.
environment	Environment	Provides environment variables that should be added to the session.

Result

Returns a result object with ID, URL, and project_id properties identifying the session

Property name	Type	Summary
id	string	The ID of the new session, used in stop_session or get_session to filter the result.
url	string	The URL starting with a / to use to join the session. If the server context has changed, prepend the appropriate https://server-name part to the returned value before opening the URL in the browser, or using it in further API requests.
project_id	string	A unique identifier for the project used by the session.

Example results:

{
  "result": {
    "url": "/s/0da29cfc78a31acdd5861/",
    "id": "acdd5861",
    "project_id": "cfc78a31"
  }
}

Method launch_job

Launches a Posit Workbench Job on behalf of the username provided in the parameters or the auth cookie. A super-admin API token is required to launch a job on behalf of a different user than the API token owner.

POST /api/launch_job {
   "method": "launch_job",
   "kwparams": {
      "job": {
         "cluster": "Local",
         "name": "Job Test",
         "exe": "/bin/bash",
         "args": ["-c", "R -f JobTest.R"]
      }
   }
}

Parameters

Name	Type	Summary
job (required)	object	An object that provides the initial properties for the job: name, exe, args, cluster, and working_directory. Optional initial properties include environment, tags, container_image, resource_limits, queues, and placement_constraints. The user property in the job can be used with a super-admin API token to impersonate another user. If the user property is not set, the user from the API token is used.
apply_config_settings	boolean	Set this to false to bypass launcher mounts and other global settings normally applied to all jobs

Result

Returns a result object with a job property containing the WorkbenchJob. The returned job has all fields populated, both the initial values provided and those that reflect the state of the running job (e.g., status, process id, the host for the job, and timestamps).

Property name	Type	Summary
job	object	The WorkbenchJob with job_id and status fields populated

Example results:

{
  "result": {
    "result": {
      "job": {
        "id": "...",
        "cluster": "Local",
        "name": "JobTest.R",
        "user": "jvroom",
        "queues": [],
        "workingDirectory": "",
        "host": "localhost",
        "status": "Pending",
        "pid": 729383,
        "command": "",
        "exe": "/bin/bash",
        "stdoutFile": "",
        "stderrFile": "",
        "stdin": "",
        "args": [
          "-c",
          "R -f /tmp/JobTest.R"
        ],
        "environment": [],
        "placementConstraints": [],
        "lastUpdateTime": "2024-10-01T20:49:32.294130Z",
        "submissionTime": "2024-10-01T20:49:32.294014Z",
        "exposedPorts": [],
        "mounts": [],
        "config": [],
        "resourceLimits": [],
        "tags": [],
        "metadata": {},
        "resourceProfile": "custom"
      }
    }
  }
}

Method get_session

Returns session and job information for a given user. An admin or super-admin API token is required to retrieve a session for a user other the API token’s owner.

POST /api/get_session {
   "method": "get_session",
   "kwparams": {
      "session_id": "id1,id2",
      "user": "username",
      "fields" : ["id", "url", "activity_state"],
      "include_jobs" : true
   }
}

Parameters

Name	Type	Summary
session_id	string	Return info for a specific session identified by the ID provided.
user	string	Return sessions for the given user, defaulting to the api token’s username. The token must have at least read-only admin or super-admin permissions with read access to get the session info for a different user.
fields	array of string	Array of field names returned by the session to filter whats returned.
include_jobs	boolean	Set to true to include job information in the response.

Result

Returns a result object with an array of sessions and optionally jobs

On error, either a 40x, 50x http status code, or a JSON RPC error will be returned with code, and message properties. Any problems associated with the user’s account, or configuration problems on the server node will generate a JSON RPC error with an appropriate message.

Property name	Type	Summary
sessions	array of WorkbenchSession
jobs	array of WorkbenchJob

Example results:

{
  "result": {
    "sessions": [
      {
        "url": "/s/0da29cfc78a316124d41f/",
        "id": "6124d41f",
        "launch_parameters": {
          "name": "RStudio Pro Session",
          "cluster": "Local",
          "placement_constraints": [],
          "resource_limits": [],
          "queues": [],
          "container_image": "",
          "default_image": "",
          "container_images": [
            ""
          ]
        },
        "created": 1721684073175,
        "activity_state": "suspended",
        "last_state_updated": 1721695376181,
        "label": "RStudio Pro Session",
        "display_name": "RStudio Pro Session",
        "status_detail": "",
        "project": null,
        "project_id": "none",
        "running": false,
        "working_dir": "~",
        "executing": false,
        "save_prompt_required": false,
        "last_used": 1721694685914,
        "r_version": "4.3.0",
        "r_version_label": "",
        "r_version_home": "/usr/lib/R",
        "suspend_size": 20324197,
        "workbench": "RStudio"
      }
    ]
  }
}

Method resume_session

Resumes a suspended Posit Workbench Session (RStudio Pro sessions only). Provide the username of the session owner and the session id. If a username parameter is not provided, the owner of the API token or auth cookie used instead. A super-admin API token is required to resume a session on behalf of a different user than the API token owner.

POST /api/resume_session {
   "method": "resume_session",
   "kwparams": {
      "username": "optional-session-username",
      "session_id": "<the-session-id-returned>",
      "launch_parameters": {
         "name": "Job name", "cluster": "Local",
         "placement_constraints":[], "resource_limits":[], "queues":[]
      }
   }
}

Parameters

Name	Type	Summary
session_id (required)	string	The session id of the suspended session to resume. Use get_session to get the session_id, or store the value returned from launch_session.
username	string	The owner of the session. If not provided the API token’s owner is used.
launch_parameters	LaunchParameters	Provides additional launch config, required when launcher sessions are enabled.

Result

Returns a result object with ID, URL, and project_id properties identifying the session

Property name	Type	Summary
url	string	The URL starting with a / to use to join the session, that was returned in launch_session.
project_id	string	The unique identifier for the project used by the session, that was returned in launch_session.

Example results:

{
  "result": {
    "url": "/s/0da29cfc78a31acdd5861/",
    "id": "acdd5861",
    "project_id": "cfc78a31"
  }
}

Method get_compute_envs

Returns metadata about the environments the server supports for launching new sessions, including resource limits for a given user. When using an admin or super-admin token with read access, it’s possible to retrieve the environment for a supplied user. Use the POST method and provide the username as a named parameter.

GET /api/get_compute_envs

POST /api/get_compute_envs {"method": "get_compute_envs", "kwparams":{ "user": "session-username"}}

Result

Returns a result object with ComputeEnv and a list of supported IDEs

Property name	Type	Summary
clusters	array of ComputeEnv
workbenches	object	An object whose properties are the names of the supported IDEs: ‘RStudio’, ‘VS Code’, ‘JupyterLab’, ‘Jupyter Notebook’

Example results:

{
  "result": {
    "clusters": [
      {
        "supportsContainers": false,
        "queues": [],
        "config": [],
        "resourceLimits": [],
        "placementConstraints": [],
        "resourceProfiles": [],
        "name": "Local",
        "type": "Local",
        "workbenches": {
          "RStudio": {
            "supports_interactive_sessions": true,
            "supports_adhoc_jobs": true
          },
          "JupyterLab": {
            "supports_interactive_sessions": true,
            "supports_adhoc_jobs": true
          },
          "Jupyter Notebook": {
            "supports_interactive_sessions": true,
            "supports_adhoc_jobs": true
          },
          "VS Code": {
            "supports_interactive_sessions": true,
            "supports_adhoc_jobs": true
          }
        },
        "images": []
      }
    ],
    "workbenches": {
      "RStudio": {
        "default_cluster": "Localhost",
        "default_image": "",
        "lang": {
          "name": "R",
          "available": true
        }
      },
      "JupyterLab": {
        "default_cluster": "Local",
        "default_image": "",
        "lang": {
          "name": "Jupyter",
          "available": true
        }
      },
      "Jupyter Notebook": {
        "default_cluster": "Local",
        "default_image": "",
        "lang": {
          "name": "Jupyter",
          "available": true
        }
      },
      "VS Code": {
        "default_cluster": "Localhost",
        "default_image": "",
        "lang": {
          "name": "code-server",
          "available": false
        }
      }
    }
  }
}

Method get_job_output

Returns a specific job’s output for either a live running job, or a job that finished but hasn’t been removed from the system yet. The output is returned in a stream of JSON objects that are parsed one after another. For a live job, the chunks are returned as they are produced and the connection will not be closed by the server until the job exits, or it times out on the server or in a proxy layer. An admin or super-admin API token with read access is required to retrieve the job output for a user other than the API token’s owner.

POST /api/get_job_output {
   "method": "get_job_output",
   "kwparams": {
      "job_id": "job_id",
      "user": "job-user"
   }
}

Parameters

Name	Type	Summary
job_id (required)	string	The id of the job.
user	string	The user that owns the job, required if using a super-admin token for impersonation.

Result

Returns a stream of JSON objects, each one containing a chunk of either the stdout or stderr streams of the job. Each chunk has jobId, outputType of stdout or stderr and output with the chunk. If the process is still running, the server will stream chunks as they arrive. If not, the stream is closed.

{"jobId":"....", "outputType":"stdout", "output":"A chunk of the job's standard output"}

{"jobId":"...", "outputType":"stderr", "output":"A chunk of the job's error output"}

Method stop_session

Stops one or more session given a list of session_ids. Optionally force quit, or suspend the sessions. Use an admin or super-admin API token with write access to stop a session on behalf of another user.

POST /api/stop_session {
   "method": "stop_session",
   "kwparams": {
      "session_ids": "id1,id2,...",
      "force_quit" : false,
      "suspend_session": false
   }
}

Parameters

Name	Type	Summary
session_ids (required)	string	A comma separated list of session IDs to stop, suspend or force quit
force_quit	boolean	Sends the kill signal to forcefully quit the session process.
suspend_session	boolean	For sessions that support suspension, e.g. RStudio, tells the session to save its state before exiting.

Result

On success, returns with an HTTP 200 return code and an empty result message.

Method stop_job

Stops a workbench job given the job_id. Use an admin or super-admin API token with write access to stop a job on behalf of another user.

POST /api/stop_job {
   "method": "stop_job",
   "kwparams": {
      "job_id": "id1"
      "force_quit" : false
   }
}

Parameters

Name	Type	Summary
job_id (required)	string	The individual job id of the job to stop.
force_quit	boolean	With force_quit is true, the kill signal is used instead of terminate (i.e. `kill -9 <pid>` instead of `kill <pid>`).

Result

On success, returns with an HTTP 200 return code and an empty result message.

Method version

Returns a JSON object with version information including a list of enabled features.

GET /api/version

Result

Returns an HTTP 401 code for permission denied, 50x for server problems, or 200 with a JSON result indicating success or error.

On success, the result is a JSON object with version and feature info.

Property name	Type	Summary
version	object	An object containing major, minor, patch, build_number, release_name, and build_type string properties
features	array of string	List of the names of features enabled in this server

Method get_users

Returns a JSON array containing the users registered in Posit Workbench. Requires read access with an admin or super-admin API token.

GET /api/get_users

Result

Returns an HTTP 401 code for permission denied, 50x for server problems, or 200 with a JSON result indicating success or error.

On success, the result is a JSON object with version and feature info.

Example results:

{
  "result": [
    {
      "username": "user1",
      "status": "Active",
      "isAdmin": false,
      "uid": 1037,
      "lastSignIn": "2024-07-30T23:45:28.351868Z",
      "created": "2024-07-30T23:41:28.109218Z"
    },
    {
      "username": "user2",
      "status": "Inactive",
      "isAdmin": false,
      "uid": 1038,
      "lastSignIn": "2024-07-25T13:43:25.231845Z",
      "created": "2024-07-30T23:41:28.109218Z"
    }
  ]
}

Data types

These types are used in the Workbench API.

PlacementConstraint

Name/values that choose launch options defined in the server configuration

Required Properties: name, value

Properties

Property name	Type	Summary
name	string	Constraint name
value	string	Constraint value

ResourceLimit

Properties

Property name	Type	Summary
name	string
value	string
valueType	string

LaunchParameters

Used to define parameters for Job launcher sessions.

Required Properties: name, cluster, resource_limits, placement_constraints, queues

Properties

Property name	Type	Summary
name	string
cluster	string	The compute environment to launch the session
container_image	string
placement_constraints	array of PlacementConstraint
resource_limits	array of ResourceLimit
queues	array of string
default_image	string
resource_profile	string
aws_role_arn	string
databricks_instance	string
snowflake_account	string
snowflake_role	string

WorkbenchSession

The metadata workbench manages for a given session.

Properties

Property name	Type	Summary
name	string
id	string
url	string
launch_parameters
created	number
activity_state	string	Overall state of the session at a given moment, for the launch process: launching/resuming, pending, starting, and finally running. Also represents exit states: finished, suspended, failed, killed, and suspending for sessions that support suspension (i.e. RStudio)
last_state_updated	number	Timestamp for last state change
label	string
display_name	string
project	string
project_id	string
working_dir	string
running	boolean
executing	boolean	The session is active and busy.
save_prompt_required	boolean
last_used	number
r_version	string
r_version_label	string
r_version_home	string
suspend_size	number
workbench	string

WorkbenchJob

The metadata workbench manages for a given session.

Properties

Property name	Type	Summary
name	string	Required job’s name.
cluster	string	The name of the compute environment to run the job - e.g. Local, Slurm, Kubernetes
exe	string	The executable on the session node to run
args	array of string	List of string arguments to provide to the exe.
command	string	An alternative to exe for providing a shell command to run (i.e. the job is run as: `/bin/sh -c command arg1 arg2...`).
working_directory	string	Provides the initial working directory for the job.
user	string	The username of the job’s owner
queues	array of string
tags	array of string
id	string	The job’s id, returned from the launch_job API.
container	string
host	string
group	string
status	string
statusMessage	string
statusCode	string
pid	number
exitCode	number
stdoutFile	string
stderrFile	string
submissionTime	string
lastUpdateTime	string

JobConfig

Properties

Property name	Type	Summary
name	string
value	string
valueType	string

ComputeEnv

An environment for running sessions.

Properties

Property name	Type	Summary
supportsContainers	boolean
queues	array of string
config	array of JobConfig
resourceLimits	array of ResourceLimit

EnvironmentVariable

Stores an environment variable.

Required Properties: name, value

Properties

Property name	Type	Summary
name	string
value	string

SessionHook

Stores information about a session hook.

Required Properties: exe

Properties

Property name	Type	Summary
exe	string

WorkbenchIDEInfo

Stores default info for a supported IDE

Required Properties: default_cluster, default_image, lang

Properties

Property name	Type	Summary
default_cluster	string
default_image	string
lang

WorkbenchLanguage

Required Properties: name, available

Properties

Property name	Type	Summary
name	string
available	boolean

WorkbenchUser

Properties

Property name	Type	Summary
username	string	Contains the username of the user.
status	string	Contains the user’s status: `Active` or `Inactive`.
isAdmin	boolean	This value is `true` if the user has the ability to access the /admin UI.
isSuperAdmin	boolean	This value is `true` if the user has the super-admin role that can also start sessions/jobs on behalf of other users.
uid	int	Contains the posix user’s `uid` used as an attribute on files for permissions.
lastSignIn	string
created	string