Transient settings migration guide

edit

Transient settings migration guide

edit

We no longer recommend using transient cluster settings. You can use transient settings to make temporary configuration changes to a cluster. However, a cluster restart or cluster instability can unexpectedly clear these settings, leading to a potentially undesired cluster configuration.

To avoid this risk, reset any transient settings you’ve configured on your cluster. Convert any transient setting you’d like to keep to a persistent setting, which persists across cluster restarts and cluster instability. You should also update any custom workflows and applications to use persistent settings instead of transient settings.

Some Elastic products may use transient settings when performing specific operations. Only reset transient settings configured by you, your users, or your custom workflows and applications.

To reset and convert transient settings:

  1. Get a list of any configured transient settings using the cluster get settings API.

    GET _cluster/settings?flat_settings=true&filter_path=transient

    The API returns transient settings in the transient object. If this object is empty, your cluster has no transient settings, and you can skip the remaining steps.

    {
      "persistent": { ... },
      "transient": {
        "cluster.indices.close.enable": "false",
        "indices.recovery.max_bytes_per_sec": "50mb"
      }
    }
  2. Copy any settings you’d like to convert into the persistent object of a cluster update settings API request. In the same request, reset any transient settings by assigning them a null value.

    PUT _cluster/settings
    {
      "persistent": {
        "cluster.indices.close.enable": false,
        "indices.recovery.max_bytes_per_sec": "50mb"
      },
      "transient": {
        "*": null
      }
    }
  3. Use the cluster get settings API to confirm your cluster has no remaining transient settings.

    GET _cluster/settings?flat_settings=true

    If the transient object is empty, your cluster has no transient settings.

    {
      "persistent": {
        "cluster.indices.close.enable": "false",
        "indices.recovery.max_bytes_per_sec": "50mb",
        ...
      },
      "transient": {
      }
    }