Hi Krystian, Quick update on next steps: I've rebooted srv04 back onto mainline v6.8.12 (the kernel that just panicked with the `migrate_swap`/`multi_cpu_stop` signature described above), this time with Viktor's suggested boot parameters added:
``` processor.max_cstate=1 idle=nomwait ``` Confirmed active via `/proc/cmdline`: ``` BOOT_IMAGE=/boot/vmlinuz-6.8.12-060812-generic root=UUID=... ro console=ttyS0,115200n8 console=tty0 vga=normal nomodeset nointremap processor.max_cstate=1 idle=nomwait ``` All 14 guest VMs are back up and running. I'll monitor this host and report back - if it survives significantly longer than the ~3 days it took to panic without these parameters (ideally without even the earlier hung-task warnings we saw on a different host), that would support Viktor's theory about idle/mwait state transitions racing with the `multi_cpu_stop`/NUMA-swap mechanism on this hardware. -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/2163642 Title: Kernel panic "Attempted to kill the idle task" on AMD Opteron multi- node NUMA under KVM (6.8.0-137) To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2163642/+subscriptions -- ubuntu-bugs mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
