We’re using FoundationDB 7.1.37 with continuous backups writing to a local mount. Our challenge is that the backup directory fills up quickly once we have two full snapshots. We run fdbbackup expire to reclaim space, but after expiration if we need older data, restoring older versions becomes impossible because the logs have been removed.
Our goals:
-
Maintain ~30 days of point-in-time restore capability on local disk
-
Retain older weeks/months for Grandfather-Father-Son style backups
-
Be able to restore from specific older points if needed (not just snapshot endVersions)
-
Avoid filling the backup disk with multiple multi-TB snapshots
We’ve read the Backup and Restore section many times, but it’s still unclear how others handle long-term retention with continuous backups while keeping restores valid.
Main issue we’re seeing
If we rsync a backup set to archive it before expiring, the archived restore sometimes fails with:
Missing backup data
FileRestoreMissingRangeFile
Trace logs show that thousands of required kvranges/snapshot.../range,* files are missing, even though fdbbackup describe reports Restorable: true.
What is a recommended approach for long-term retention with continuous backups in FoundationDB?