VM Migration between nodes crashes node

fabian said:
could you maybe also post the VM config? thanks!
Click to expand...
Sure, it isn't unique to a VM, I can create any VM and have the same problem with it.

Code: 
root@node5:~# qm config 102
agent: 1
boot: order=scsi0;ide2;net0
cores: 4
cpu: host
ide2: none,media=cdrom
memory: 32768
meta: creation-qemu=6.2.0,ctime=1660494772
name: stress-1
net0: virtio=82:77:B4:3B:11:CF,bridge=vmbr601,firewall=1
numa: 0
ostype: l26
scsi0: node5-tank:vm-102-disk-0,cache=writeback,format=raw,iothread=1,size=100G
scsihw: virtio-scsi-pci
smbios1: uuid=d08d0386-d14c-47a4-8e05-3680d015ea1b
sockets: 1
vmgenid: c6c860f9-6236-45a5-8b86-e41f69bf7091
root@node5:~#
 
fabian said:
so nothing in the journal -> next step would be to dump the actual console (via IPMI/netconsole/serial console) in the hopes that something gets logged there.. it's strange that the node reboots on its own though, usually you'd expect a kernel bug to print some sort of trace and hang :-/
Click to expand...
Agree, I'm thinking hardware issues, but it only happens during the migration. I ran the VM on the "bad" nodes under stress for cpu/ram/disk and it survived 24 hours under stress.

Yet a migration causes the reboots.
 
could you try with cpu type kvm64 instead of host?
 
worth a shot. then the only thing that remains is to see whether it spits out any message on the local console (/netconsole/serial console) when it crashes.
 
fabian said:
worth a shot. then the only thing that remains is to see whether it spits out any message on the local console (/netconsole/serial console) when it crashes.
Click to expand...
Swapped one of the bad nodes, so far no crashes after 50 or so migrations to and from it.
 
  • Like
Reactions: fabian
You must log in or register to reply here.
Top Bottom