Saturday I built a new NAS and I gave it the HWE ubuntu 20.04 kernel (5.13) because newer is better right?
With kernel 5.13.0-39 I can start VMs but they will hang and my kernel logs fill up with:
Hardware is a Supermicro X10SDV-TLN4F with a Intel(R) Xeon(R) CPU D-1540 @ 2.00GHz
I somehow got the idea to try and go back to the GA kernel (5.4) and hooray with 5.4 VMs start and run just fine.
Now this morning I had rebooted the first node in my 3-node LXD cluster and that node also upgraded to the 5.13.0-39-generic kernel. Unfortunately this left me with the exact same kernel messages.
This cluster member hardware is a X10SRM-TF with a Intel(R) Xeon(R) CPU E5-2683 v4 @ 2.10GHz
Both my standalone NAS and my have different motherboard and CPU…
At first I thought it was a problem with my NAS’s board/cpu but now that my other node with completely different hardware has the exact same issue with 5.13.0-39 I started debugging.
I checked my prometheus metrics and saw that this morning my node1 ran with kernel 5.13.0-35 so I just restarted my node1 and selected the -35 kernel in grub and with -35 I can start VMs just fine. Then I tried -37 and that has the same issues as -39!
While typing this message I reported this to the Ubuntu Security team. I do not expect this issue to be related to LXD but rather the new Intel security fixes that were merged on March 31 and April 1st.
So I’m posting this here incase anyone else runs into this issue or has a pending reboot for their HWE kernel