RDI instance settings

Change the settings of an RDI instance while it runs

Instance settings control the behavior of a single RDI instance. You can change them while RDI runs, without a Helm upgrade or a reinstallation, and changing them does not restart the RDI operator or the RDI API. A setting that you set overrides the corresponding installation value until you delete it.

Each instance has its own settings, even when two instances share the same RDI database in a high availability setup. Run the commands, or send the API requests, to the instance that you want to change.

Manage instance settings

Use the following CLI commands to manage the settings of an instance:

# List all the settings and their values
redis-di list-settings

# Show the value of one setting
redis-di get-setting standby_mode

# Set a setting
redis-di set-setting standby_mode true

# Delete a setting, so that the installation value applies again
redis-di delete-setting standby_mode

The CLI shows a setting that is not set as <none>. See list-settings, get-setting, set-setting, and delete-setting for the command options.

With RDI API v2, GET /api/v2/settings returns all the settings of the instance, with null for a setting that is not set. PATCH /api/v2/settings applies its body as a JSON merge patch: a value sets a setting, null deletes it, and the settings that the body omits are unchanged. For example, the following body sets standby_mode:

{"standby_mode": true}

Available settings

Setting Type When not set
standby_mode Boolean The operator.leaderElection.standbyMode Helm value, false by default

standby_mode

Controls whether the instance takes part in the leader election that decides which instance runs the pipelines. Use it to fail over to another instance deliberately, for example before maintenance. See Trigger a failover manually for the steps.

  • true: The instance leaves the leader election. If it is the leader, it gives up the leadership, so that another instance can take over. Its pipelines change to the standby status, and RDI removes their collectors and stream processors from it.
  • false: The instance takes part in the leader election, even if operator.leaderElection.standbyMode is true in its Helm values.
  • Not set: The operator.leaderElection.standbyMode Helm value applies.

The operator applies a change on its next leader election check, within 10 seconds with the default configuration. The setting has no effect while leader election is disabled.

RATE THIS PAGE
Back to top ↑