Upgrades, Maintenance, and Disaster Recovery
Domain 7 - Upgrades, Maintenance, and Disaster Recovery
This domain tests lifecycle work: upgrade sequencing, routine maintenance that keeps copies healthy, and recovery of the CommCell itself when the CommServe is lost.
Core rules
- Upgrade order is CommServe first, then MediaAgents, then clients. Running a newer MediaAgent against an older CommServe is unsupported, so check the compatibility matrix before pushing software.
- Before any upgrade, run a CommServe disaster recovery backup and confirm it completed to a location off the CommServe host. That backup contains the CommCell database and configuration and is the foundation of any rebuild.
- Routine maintenance includes deduplication database verification, media recycling and aging, library inventory and drive cleaning, and pruning copies whose retention has expired.
- A failed dedupe database does not mean data loss: the database can be restored from its own backup or reconstructed from the data store, which takes time and I/O.
- CommServe DR follows a fixed sequence: install a CommServe of a compatible version on the new host, restore the latest DR backup, verify clients and storage policies, then reinstall or reconnect MediaAgents. Index caches are not part of the DR backup and are rebuilt or restored separately.
- Client and MediaAgent reinstalls should preserve identity so existing backups and policies remain associated.
Worked example
The CommServe host is destroyed. Build a replacement server with the same name and a supported version, install the CommServe software without creating a new CommCell, then restore the most recent DR backup. After the database is restored, log in and confirm clients, storage policies and schedules are present. Reinstall MediaAgents pointing at the restored CommServe, then validate by browsing and test-restoring a recent job. Only after that should new backups be enabled, so a bad restore is not overwritten by fresh data.
Traps that cost marks
- Upgrading MediaAgents before the CommServe.
- Skipping the pre-upgrade DR backup because the upgrade "usually works".
- Assuming the DR backup includes MediaAgent index caches.
- Starting new backups immediately after a CommServe restore before validating that restored jobs are browsable.