Linux Kernel bug 19.2.1 squid

[tcrawford22]

We were running rook ceph on Ubuntu 20.04 kernel version 5.4 this is on our k8s cluster, however, see further along this seems to be a issue with cephfs not rook
Because upon updating to Ubuntu 22.04, kernel version 5.15.0-136 we started getting errors causing reboots

kernel: BUG: kernel NULL pointer dereference, address: 0000000000000402
kernel: #PF: supervisor read access in kernel mode
kernel: #PF: error_code(0x0000) - not-present page

Looking around we found the following posts all pointing the page fault at ceph
https://forum.proxmox.com/threads/bug-kernel-null-pointer-dereference-address-0000000000000402.106067/
https://access.redhat.com/solutions/6993035
https://lkml.iu.edu/2205.0/04018.html
so we updated to
5.15.178-0515178-generic

However, the crashes are still occurring.
Any advice is appreciated

[Nikita-baghela07]

@tcrawford22
The crashes you’re seeing are due to a known kernel bug in the CephFS client code on Linux 5.15, which causes NULL pointer dereferences during certain Ceph operations. The fix is not in the 5.15.x series you’re running; upgrading to a newer kernel (5.19+, or ideally 6.1 LTS or 6.8/6.9) is the most reliable solution.

Plan for You:

Upgrade Ubuntu 22.04 nodes to HWE kernel (6.8.x).

Test CephFS mounts again; this should eliminate the NULL pointer dereference crashes.

If upgrading the kernel is not immediately possible, switch to ceph-fuse for mounting until you can move off 5.15.