[SOLVED] Garbage collection error "unexpected error on datastore traversal" appears regulary

[antipiot]

Hello!

I have a Proxmox Backup Server running for a while and sometimes encounter this error:
TASK ERROR: unexpected error on datastore traversal: No such file or directory (os error 2)
GC is scheduled to run daily.



Proxmox Backup server is in a VM on a Proxmox VE host (not the one who's backuped )
Datastore is on a zfs pool

Any idea why?

Regards!

JS

 

[t.lamprecht]

Hi,

I have a Proxmox Backup Server running for a while and sometimes encounter this error:

Did you update it frequently too, i.e., what version are you currently on?

Any idea why?

Does a prune job runs during GC? In early versions of Proxmox Backup Server the GC would error out if a backup index vanished during a run, but that was fixed quite a bit ago.

 

[antipiot]

Hi! Thanks for your fast answer!

Hi,

Did you update it frequently too, i.e., what version are you currently on?

Yes, the servers are up to date: Backup Server 1.0-11

Does a prune job runs during GC? In early versions of Proxmox Backup Server the GC would error out if a backup index vanished during a run, but that was fixed quite a bit ago.

Seems they are yes, i will change the run times to mitigate this following what you said.



Regards

 

[t.lamprecht]

Yes, the servers are up to date: Backup Server 1.0-1

That should be at least 1.0-11 though? If not, ensure you got a valid repo you can access setup and update to latest:
https://pbs.proxmox.com/docs/installation.html#debian-package-repositories

 

[antipiot]

That should be at least 1.0-11 though? If not, ensure you got a valid repo you can access setup and update to latest:
https://pbs.proxmox.com/docs/installation.html#debian-package-repositories

It is indeed - bad copy/pasta - sorry

 

[t.lamprecht]

hmm, OK... But yeah, then I'd for now set the pruning schedule a bit earlier than GC, 10 minutes to half an hour should be easily enough. That way GC has a slightly better chance to clean up more data.

 

[antipiot]

hmm, OK... But yeah, then I'd for now set the pruning schedule a bit earlier than GC, 10 minutes to half an hour should be easily enough. That way GC has a slightly better chance to clean up more data.

Makes sense! will go for:


Thanks for your help!

 

[ahoi_]

Got the same error tonight:

2021-05-14T00:00:00+02:00: starting garbage collection on store local-storage
2021-05-14T00:00:00+02:00: task triggered by schedule 'daily'
2021-05-14T00:00:00+02:00: Start GC phase1 (mark used chunks)
2021-05-14T00:00:00+02:00: TASK ERROR: unexpected error on datastore traversal: No such file or directory (os error 2) - "/storage/vm/101/2021-05-05T21:00:01Z"

So maybe the default settings should be adapted or there should be a hint in the docs?

Cheers

 

[alfredo.pironti]

+1 got the same error. It would indeed seem useful to update the defaults. Also, as of now, if you don't pick the "daily at 0:00" you can only pick the "daily at 2:30; 22:30" which actually seems to run the task (either prune or GC) twice. Having a "daily at 22:30" option by default would likely put this issue to sleep?

 

[Herman]

I ran into the same issue the past couple of days, and noticed that I can just put a time value in the box, so 04:00 instead of daily, and that seems to work.