Environment

• YugabyteDB Anywhere - 2.18+ installed using YBA Installer

Issue

One or more Universe backups are skipped when a backup schedule conflicts with YugabyteDB Anywhere (YBA) backups created using yba-ctl createBackup. Messages similar to the following appear in the YBA application.log (default location: /opt/yugabyte/data/logs/application.log):

YW 2024-02-05T20:00:02.247Z [INFO] 3c32f135-1234-5678-9abc-abef183205ea from HikariCPConnectionPool in ForkJoinPool.commonPool-worker-5 - Shutting down connection pool. 
YW 2024-02-05T20:00:02.252Z [INFO]  from AkkaHttpServer in application-akka.actor.internal-dispatcher-128462 - Running provided shutdown stop hooks
YW 2024-02-05T20:15:33.539Z [INFO]  from PerfAdvisorDB in main - Database perf_advisor already exists

Note the shutdown messages followed by the 15 minute gap in the logs. This sequence of messages indicates that the YBA service was shut down and restarted.

Resolution

Overview

The default behaviour for yba-ctl createBackup shuts down the YBA software during YBA backups, so Universe backups that are scheduled to run at this time will be skipped.

Workaround

Reschedule the YBA backups to a time when Universe backups are not expected to run.

Permanent Fix

Add the --skip_restart flag to the yba-ctl createBackup command being used to create the YBA backups. For example:

/usr/bin/yba-ctl createBackup --skip_restart

Additional Information

Jira ticket PLAT-12912 is open to make skip restart the default behaviour in a future release.