Possible datastore corruption after download canceled.

[EllyMae]

PBS 2.2-7 running in a VM on Proxmox 7.2-11.

I was doing a file download on Proxmox VE a directory of files to a .zip file. Downloading from a PBS datastore. I changed my mind about something and canceled the download on the browser. Now I can't download anything from PBS.



I've restarted PBS, but that doesn't help. Trying to avoid restarting the whole Proxmox server. Currently running a verify on the whole datastore, but that's going to take quite a while. Restoring an image file from the PBS GUI seems to be working ok. However, I really want to get just a certain directory.

Any ideas as to why this timeout is happening or how to recover from it?
Thanks.

 

[Moayad]

Hello,

Is anything interesting in the Syslog from Proxmox VE side?

 

[EllyMae]

Hello,

Is anything interesting in the Syslog from Proxmox VE side?

Yes! There are many, many of these, and they are still happening.
Nov 15 11:01:58 proxmox QEMU[1917139]: error during PBS read: command error: broken pipe

 

[EllyMae]

I restarted Proxmox VE. Now it's working again. Did I run into a bug?

 

[Moayad]

Hello,

In general, the broken pipe is caused by an Input/Output Error. In your case, I would check the monitor on your PVE in order to see if there was a high IO at the problem time.

 

[MathieuL]

Hello,
coming accross this thread while searching why my logs are fulled with

error during PBS read: command error: broken pipe

And in fact I had a file restore cancelled days ago too

Find out that the restore process was still running and killed it solved the problem, in case it helps someone :

ps aux | grep restore

returns me a process with qemu-system-x86_64 within I see args with the vm previously cancelled (via GUI), so just killed it and everything goes back to normal

 

[ciniset]

Hello,
coming accross this thread while searching why my logs are fulled with

error during PBS read: command error: broken pipe

And in fact I had a file restore cancelled days ago too

Find out that the restore process was still running and killed it solved the problem, in case it helps someone :

ps aux | grep restore

returns me a process with qemu-system-x86_64 within I see args with the vm previously cancelled (via GUI), so just killed it and everything goes back to normal

Thank you so much! It really save me !