Backups failing back to Full Backups

acebmxer

So not sure why this problem has come back but all backups are failing back to Full backups.

2026-07-06T02_09_34.032Z - backup NG.txt <- this was a Planned full backup.

2026-07-07T02_00_00.007Z - backup NG.txt - first Delta backup after full fell back to full.

I just tried to disable "Purge snapshot data when using CBT" but still fell back to full. Unless i need to run a nother full job first?

poddingue

I think you've hit the known purge-snapshot-data issue. The XO incremental backup docs have a "Known issues" note (https://docs.xen-orchestra.com/xo5/incremental_backups#known-issues) about this: with purge snapshot data on, backups can transfer a full on every run, and the two suggested fixes are to disable purge (which you tried) or migrate the disk to another storage to reset the disk state.
Since disabling purge didn't clear it, the disk migration is probably the one that resets the CBT chain.
There's also an open issue matching your "it came back" symptom (https://github.com/vatesfr/xen-orchestra/issues/8713) where leftover invalid CBT snapshots keep forcing fulls until the chain is reset, so it's already tracked.
I could be wrong about which step actually clears it, so if migrating the disk fixes it for you, dropping a note on that issue would help the next person.

rtjdamen created this issue in vatesfr/xen-orchestra

open CBT backup fails repeatedly due to leftover invalid CBT snapshots #8713

acebmxer

@poddingue

while looking into this further i thought I might be facing disk images showing up a snapshots because i saw this when i looked into the storage. Just an update on what may have cause my fall back to full backups if its releated. I migrated my 2 host from AMD to Intel I kinda did it a sloppy way. What I mean by that, is I think I migrated them between two pools more then need as i had 3 host already setup as intel. In the end they made their way back to the original SR after AMD host were removed from the picture. The first backup after migration was I believe a Full backup as it was over the weekend.

After expanding I saw my actual vms still there... I think those are from the current backups? since i truned off remove snapshot when using cbt as that used to fix this issue. But maybe it hasnt as the vdi's still cbt enabled?

acebmxer

After removing the CBT enabled option on the actual vms and ran a delta backup it did not fall back to full.

2026-07-08T09_59_01.627Z - backup NG.txt

These are still here....

And there is this message about those files.

acebmxer

I re-enabled purge snapshot data when using CBT. And it did not fall back to full again either.....

2026-07-08T10_05_49.749Z - backup NG.txt

The unknow snapshot files are still pressent.

acebmxer

@poddingue Just ran one more backup to confirm working, and it fell back to full backup again....

What logs or other information should I look into to find the root cause or do you still think it fall under the known issue?