Link Search Menu Expand Document Documentation Menu
Documentation

Get ML Task API

You can retrieve information about a machine learning (ML) task (such as model training, deployment, or prediction tasks) using the task_id.

This API is different from the general Tasks API, which tracks general OpenSearch operations and has a different response format.

Endpoints

GET /_plugins/_ml/tasks/<task_id>

Example request

GET /_plugins/_ml/tasks/MsBi1YsB0jLkkocYjD5f

Example responses

The response format depends on the task state. Different ML operations (for example, model training, deployment, or registration) return different response formats based on their current status.

Task in progress

While a task is still running, the response includes task details but excludes the model_id:

{
  "task_type": "DEPLOY_MODEL",
  "function_name": "TEXT_EMBEDDING",
  "state": "CREATED",
  "worker_node": ["KfEEGG7_SsKZVFqI4ko2FA"],
  "create_time": 1767030135146,
  "last_update_time": 1767030135771,
  "is_async": true
}

Task completed

When a task completes successfully, the response includes the model_id and full task details:

Model deployment task:

{
  "model_id": "Qr1YbogBYOqeeqR7sI9L",
  "task_type": "DEPLOY_MODEL",
  "function_name": "TEXT_EMBEDDING",
  "state": "COMPLETED",
  "worker_node": [
    "N77RInqjTSq_UaLh1k0BUg"
  ],
  "create_time": 1685478486057,
  "last_update_time": 1685478491090,
  "is_async": true
}

Model registration task:

{
  "model_id": "aVeif4oB5Vm0Tdw8zYO2",
  "task_type": "REGISTER_MODEL",
  "function_name": "TEXT_EMBEDDING",
  "state": "COMPLETED",
  "worker_node": [
    "4p6FVOmJRtu3wehDD74hzQ"
  ],
  "create_time": 1694358489722,
  "last_update_time": 1694358499139,
  "is_async": true
}

Model training task:

{
  "model_id" : "l7lamX8BO5w8y8Ra2oty",
  "task_type" : "TRAINING",
  "function_name" : "KMEANS",
  "state" : "COMPLETED",
  "input_type" : "SEARCH_QUERY",
  "worker_node" : "54xOe0w8Qjyze00UuLDfdA",
  "create_time" : 1647545342556,
  "last_update_time" : 1647545342587,
  "is_async" : true
}

Response fields

The following table lists all response body fields.

Field Data type Description
model_id String The unique identifier for the ML model. Available when task is completed.
task_type String The type of ML operation (for example, REGISTER_MODEL, DEPLOY_MODEL, or TRAINING).
function_name String The ML function type (for example, TEXT_EMBEDDING or KMEANS).
state String The current task state. Valid values are CREATED (task created), RUNNING (task actively executing), COMPLETED (task finished successfully), FAILED (task encountered an error), CANCELLED (task was canceled), COMPLETED_WITH_ERROR (task completed with errors), CANCELLING (task being canceled), EXPIRED (task expired), and UNREACHABLE (task node unreachable).
worker_node Array An array of node IDs of nodes on which the task is running.
create_time Long The timestamp when the task was created, in milliseconds since epoch.
last_update_time Long The timestamp of the last status update, in milliseconds since epoch.
is_async Boolean Whether the task runs asynchronously. Usually true for ML tasks.
input_type String The input type for training tasks (for example, SEARCH_QUERY).
350 characters left

Have a question? .

Want to contribute? or .