Hi Dagster folks! When the daemon tries to dispatch a job but is unable to do so due to the user code server being unavailable (e.g. network blip), it marks the job as failed. Is there a way to configure the daemon to retry a few seconds/minutes later, optionally with an exponential backoff?

We do have this - check out the config schema on QueuedRunCoordinator here (the last two in the list): https://docs.dagster.io/_apidocs/internals#dagster._core.run_coordinator.QueuedRunCoordinator

Oh

max_user_code_failure_retries

!