Sounds serious. I hope you can capture the net traffic causing the issue, I'm not much an expert on that.
I have crashed host though, not lately, always hardware related.
In this case what is the method providing eth? Bios enabled VT srvio, promiscuous l1 modes, vlans, macvtaps, dedicated physical vfio? That's worst to best...
Also, the machine type may be relevant, defined in the vm's xml.
<emulator>/usr/bin/kvm</emulator>???
<type arch='x86_64' machine='pc-q35-???'>hvm</type>
If the host rebooted then a watchdog or something was alive, there may be some logging of a shutdown on the host. And you do mean a reboot and not a DE logoff/logon cycle? I've seen both of those, in addition to the last crash type where the watchdog eats the canary and dies requiring a manual power cycle.
I have crashed host though, not lately, always hardware related.
In this case what is the method providing eth? Bios enabled VT srvio, promiscuous l1 modes, vlans, macvtaps, dedicated physical vfio? That's worst to best...
Also, the machine type may be relevant, defined in the vm's xml.
<emulator>/usr/bin/kvm</emulator>???
<type arch='x86_64' machine='pc-q35-???'>hvm</type>
If the host rebooted then a watchdog or something was alive, there may be some logging of a shutdown on the host. And you do mean a reboot and not a DE logoff/logon cycle? I've seen both of those, in addition to the last crash type where the watchdog eats the canary and dies requiring a manual power cycle.
Statistics: Posted by CwF — 2024-01-18 15:51
