[SOLVED] /dev/mapper/pve-root filling up on file transfer

[noviceiii]

Dear all

Moved from ESXi to proxmox I already gain experiences

I have an issue with /dev/mapper/pve-root running out of disk space. It does that until no space is left.
It does this, if several 1 TB files (database backup, zip files) with SCP (in one go) from one vm to another.
I have extended pve_root to buy me some time. It now has 50GB which should be plenty IMHO.

Further, I have set the disc controller to virtIO, changed to direct write (unsafe) and disabled syslog. That bought some time too, but neither action avoids the root partition of the host to be filled up.

Is there a way to prevent the guest systems filling up the hosts root partition?

Kind regards,
Oliver

df -h

 

[noviceiii]

Digging a bit further into it, I may face two issues:
- one is the mentioned fillup of the pve root
- second is the IO delay by the discs

For the latter I suspect a crappy onboard SATA controller on the mainboard. I may just have to life with that.
For the first, any advice would be welcome.

As a work around, one can limit the bandwith of scp by adding -l flag.
E.g, this limits the speed to 1 MB/s

scp -l 10000 /source/file/ user@server:/target/dir/

 

[Dunuin]

First you should find out what exactly fills up your pve-root. You could try something like du -a / | sort -n -r | head -n 20 to see what are the biggest files and folders.

 

[noviceiii]

Indeed a rather obvious suggestion..

It seems that somebody (me) had messed up fstab. Not juts a little. Realy. I had some false mounts that saved data to the /mnt directory.

Thank you.