Zurab Kiknavelidze
06/17/2026, 6:20 AMRocky
06/17/2026, 6:21 AMautomated_cluster_snapshot_interval_seconds is 600 seconds (10 minutes). Increasing it to 1 hour is supported.
* Shared-data Clusters Only: This specific "Automated Cluster Snapshot" feature is currently only available for shared-data clusters.
* Trade-off (RPO): The primary consideration is your Recovery Point Objective (RPO). StarRocks only retains the single latest automated snapshot. If you set the interval to 1 hour, your "worst-case" data/metadata loss (in terms of state recovery) would be 1 hour if you had to restore the cluster from object storage.
* Resource Impact: A longer interval (like 1 hour) reduces the frequency of snapshot tasks, which can slightly reduce the overhead on your FE and object storage, though the process is generally efficient.
How to configure it
You can update the interval using the following command on the FE:
sql
-- Set the interval to 1 hour (3600 seconds)
ADMIN SET FRONTEND CONFIG ("automated_cluster_snapshot_interval_seconds" = "3600");
-- Ensure the feature is enabled
ADMIN SET AUTOMATED CLUSTER SNAPSHOT ON;
Limitations to Keep in Mind
* Single Snapshot: The system automatically drops the previous automated snapshot as soon as a new one is created. You cannot maintain a history of multiple automated snapshots using this parameter.
* Beta Status: This feature is currently in Beta. It is recommended to verify your recovery process in a test environment.
If you are using a shared-nothing cluster, this specific automated feature does not apply; you would instead use manual BACKUP tasks, which can be scheduled via external tools (like cron or Airflow).
References
• administration/cluster_snapshot.md
• administration/management/FE_configuration.md
• administration/management/BE_configuration.md