Undeploy Model API

To undeploy a model from memory, use the undeploy operation.

For information about user access for this API, see Model access control considerations.

Endpoints

POST /_plugins/_ml/models/{model_id}/_undeploy
POST /_plugins/_ml/models/_undeploy

Path parameters

The following table lists the available path parameters.

Request body fields

The following table lists the available request body fields.

Example request: Undeploying a model from all ML nodes

POST /_plugins/_ml/models/MGqJhYMBbbh0ushjm8p_/_undeploy

Example request: Undeploying specific models from specific nodes

POST /_plugins/_ml/models/_undeploy
{
  "node_ids": ["sv7-3CbwQW-4PiIsDOfLxQ"],
  "model_ids": ["KDo2ZYQB-v9VEDwdjkZ4"]
}

Example request: Undeploying specific models from all nodes

POST /_plugins/_ml/models/_undeploy
{
  "model_ids": ["KDo2ZYQB-v9VEDwdjkZ4"]
}

Example response

{
  "sv7-3CbwQW-4PiIsDOfLxQ" : {
    "stats" : {
      "KDo2ZYQB-v9VEDwdjkZ4" : "undeployed"
    }
  }
}

If a model is not deployed on a node, the status for that model is not_found.

Automatically undeploy a model based on TTL

Models can be automatically undeployed from memory based on the predefined time-to-live (TTL) when the model was last accessed or used. To define a TTL that automatically undeploys a model, include the following ModelDeploySetting in your machine learning (ML) model. Note that model TTLs are checked periodically by a sync_up cron job, so the maximum time that a model lives in memory could be TTL + the sync_up job interval. The default cron job interval is 10 seconds. To update the cron job internally, use the following cluster setting:

PUT /_cluster/settings
{
    "persistent": {
        "plugins.ml_commons.sync_up_job_interval_in_seconds": 10
    }
}

Example request: Creating a model with a TTL

POST /_plugins/_ml/models/_register
 {
   "name": "Sample Model Name",
   "function_name": "remote",
   "description": "test model",
   "connector_id": "-g1nOo8BOaAC5MIJ3_4R",
   "deploy_setting": {"model_ttl_minutes": 100}
 }

Example request: Updating a model with a TTL when the model is undeployed

PUT /_plugins/_ml/models/COj7K48BZzNMh1sWedLK
{
    "deploy_setting": {"model_ttl_minutes" : 100}
}