By default, new and undeployed pipelines are set at the draft service level. Undeployed production pipelines also become draft pipelines. When you deploy a pipeline, it becomes a development pipeline unless you explicitly set its service level to production.

To learn more about service levels, see About Pipelines.

You can change the pipeline’s service level from the deepset Cloud interface or through an API endpoint.

Prerequisites

A pipeline must be deployed to become a production pipeline.

Change the Service Level from the UI

1. In deepset Cloud, go to Pipelines.
2. Find the pipeline whose service level you want to change and click its name. You land on the pipeline details page.
3. Do one of the following:
   1. Switch the production toggle to change the pipeline’s service level to production. When you do this, the pipeline’s production hours start being metered.
   2. Switch the production toggle off without undeploying the pipeline to change its service level to development. When you do this, development hours start being metered for this pipeline.
   3. Undeploy the pipeline to change its service level to draft. No hours are metered for a draft pipeline.

Change the Service Level Through REST API

• To set the service level to development or production, use the Update Pipeline endpoint.
• To set the pipeline to draft, undeploy it using the Undeploy Pipeline endpoint.

Change to Production or Development

Follow the step-by-step explanation of the code:

Or copy this code, replacing the workspace, pipeline name, and API key:

curl --request PATCH \
     --url https://api.cloud.deepset.ai/api/v1/workspaces/<YOUR_WORKSPACE>/pipelines/<YOUR_PIPELINE> \
     --header 'accept: application/json' \
     --header 'authorization: <API_KEY>' \
     --header 'content-type: application/json' \
     --data '
{
  "service_level": "DEVELOPMENT" # to change to production, replace with "PRODUCTION"
}
'

For more details, see Update Pipeline.

Change to Draft

Undeploying a pipeline changes its service level to draft.

Follow the walkthrough of the code:

or copy this code, replacing the workspace, pipeline name, and API key:

curl --request POST \
     --url https://api.cloud.deepset.ai/api/v1/workspaces/<YOUR_WORKSPACE>/pipelines/<YOUR_PIPELINE>/undeploy \
     --header 'accept: application/json' \
     --header 'authorization: <YOUR_API_KEY>'