PBS 4.0.21 /run inode exhaustion (819200)

[vbaz]

Hello,

I'm running PBS 4.0.21 under PVE 9.1.1.

I've a local datastore and a remote S3 datastore with a sync job

I noticed after a few hours / days , the PBS lxc is becoming unresponsive due to inode exhaustion for /run

The max number of inodes being : 819200

How to fix the issue ?

thx

 

[cheiss]

Hi,

see also this forum thread: https://forum.proxmox.com/threads/pbs-4-0-20-s3-cache-becomes-unavailable.176228/, seems to be the same symptoms/problem.

This is actually already fixed & available with 4.0.22 [0], so updating to that should fix it.

[0] https://git.proxmox.com/?p=proxmox-...3;hp=212fad32672bae83b21cb132492c90c88714330c

Edit: fixed the forum link

 

[t.lamprecht]

How to fix the issue ?

This should be fixed with proxmox-backup-server version 4.0.22-1, currently available on pbs-test and soon to be moved to the pbs-no-subscription.

As a stop-gap until then you should be able to use: mount -o remount,nr_inodes=0,inode64 /tmp

 

[vbaz]

This should be fixed with proxmox-backup-server version 4.0.22-1, currently available on pbs-test and soon to be moved to the pbs-no-subscription.

As a stop-gap until then you should be able to use: mount -o remount,nr_inodes=0,inode64 /tmp

Thx a lot

I'll try

Regards

 

[Chris]

To add to the responses provided, make sure to reboot the system after installing the upgraded 4.0.22-1 package, as the systemd mount unit will only take effect after that.

 

[t.lamprecht]

Or just update the inode limit for /tmp until you get around until the next reboot:

As a stop-gap until then you should be able to use: mount -o remount,nr_inodes=0,inode64 /tmp

And this only affects S3 backed datastores.