12 days uptime on a single host is not a negative proof for a low- frequency race condition - on 6.8 it also took 5 to 14 days to trigger under specific memory pressure.
Between 6.8 and 6.13, NUMA balancing heuristics and scheduler preemption changed significantly, altering the timing and frequency of migrate_swap() calls under KVM. The call trace in comment #27 caught the stopper thread directly stuck in migrate_swap(), which is precisely what 009836b4fa52 and c861cac950fb fix. Backporting both to 6.8 LTS remains the right and safe move. -- 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
