Creating or destroying Xen domU crashes the entire host
Ugh, this is not good. I have multiple Xen servers, different hardware but all Alpine 3.16 x86_64. Sunday I noticed that one of them was unresponsive, and this have never happened before, usually they have months of uptime before maintenance requires rebooting. After arriving at the site, I could conclude that then entire server was freezed, not even Magic SysRq key sequences had any effect. I did a hard reboot and tried creating a domU, which worked without problems. Monday, the server crashed again, this time at the exact moment I created a domU. I decided to check for hardware problems.
Now, the other servers with completely different hardware has also started to crash in the same way, so this problem cannot be related to hardware.
All servers are Alpine 3.16 x86_64 with latest updates as of 10/08-22, however some run kernel 5.15.57 as they are installed in "Diskless Mode". All domUs are PVH, mostly Alpine from edge to oldest supported release.
The servers does not always crash when creating or destroying a domU, and if the domU was successfully created, it will run without problems. I have tried high load and network traffic for 24 hours without problems, but as soon as the domU was rebooted, entire host crashed. The servers are also stable, if just booting dom0 and doing stuff without creating domUs.
If I create an Alpine domU on one of the servers, with a script in local.d that reboots after 10 seconds, the entire host will crash reliable within 60 seconds.
Before Sunday, no problems whatsoever, so this must be something newly added to 3.16 repo. I don't belive the kernel is the problem.
Can anyone help?
Thank you.
Regards, Mogens Jensen