VM as readonly. Why?

[iruindegi]

Hi,
This is our setup:
3 Servers with proxmox 6.3.3 in HA and Ceph. For Ceph's corosync we have a dedicated VLAN and 10G switch.
All our VM disks are in the ceph shared.

The problem is that suddenly some vm-s detect filesystem errors and mount it as readonly. We have to manually fsck the disk to fix errors and reboot. After that the VM works properly.The problem is that not always is the same VM and we don't know why is happening that.
Any clue?
Ceph monitor is in green color

 

[iruindegi]

this is the output of fsck /dev/vda2



Two days in a row happening the same. any help with this¿?

 

[tcit9880]

We are also experiencing the same issue on our 3-node ceph cluster using a dedicated 10G switch.
pve-manager/8.1.4/ec5affc9e41f1d79 (running kernel: 6.5.13-1-pve)

Did you manage to pinpoint the cause of this issue iruindegi?