Good way to fix "unable to access non-existent chunk" after changing datastore

[Senin]

Hi!

I used temporary datastore for initial setup of my PBS.
Now I've added production datastore and switched existing tasks to this new datastore.
Im receiving error for garbage collection job on a new datastore ("unable to access non-existent chunk" in the log).
I think that it's because part of the backup is still on old temp datastore because there is no such error for VMs that were initially backed up to new datastore.

The question is what is the good way to fix this?
Delete old backups from old datastore?
Delete backup job and create it again?
Sync it in some way?

Thanks for help, Alex.

 

[dcsapak]

it sounds there is something wrong, making backups on a different datastore should not skip any chunks at all.

can you maybe post your datastore config and the gc log ?

 

[Senin]

Hi!

Not sure about datastore config,
less /etc/proxmox-backup/datastore.cfg shows this

datastore: ru1vol1
comment
gc-schedule daily
path /mnt/ru1vol1
prune-schedule daily

datastore: ru1vol2
comment
gc-schedule daily
path /mnt/ru1vol2
prune-schedule daily

datastore: ru1vol3
comment
gc-schedule daily
path /mnt/ru1vol3
prune-schedule daily


vol1 & vol2 are temp datastores connected to IBM V3700 via iscsi on dedicated network (no issues with this)
vol3 is production datastore connected to QNAP via iscsi on dedicated network

Log of GC job is in attached file (some records were deleted because of large size of the file)

 

[Senin]

ok, I just removed all bad backups and removed backup job for VMs that have problems
After creating new backup job everything works fine