diff options
Diffstat (limited to 'results/classifier/gemma3:12b/kvm/2006')
| -rw-r--r-- | results/classifier/gemma3:12b/kvm/2006 | 43 |
1 files changed, 43 insertions, 0 deletions
diff --git a/results/classifier/gemma3:12b/kvm/2006 b/results/classifier/gemma3:12b/kvm/2006 new file mode 100644 index 00000000..093c87d0 --- /dev/null +++ b/results/classifier/gemma3:12b/kvm/2006 @@ -0,0 +1,43 @@ + +migrating failed with rcu_preempt message on proxmox 8 +Description of problem: +when i migrate the VM from one host to another, it fails and give messages: + + ``` +[ 584.109502] rcu: INFO: rcu_preempt detected stalls on CPUs/tasks: +[ 584.109534] rcu: 1-...!: (0 ticks this GP) idle=1408/0/0x0 softirq=8428/8428 fqs=0 (false positive?) +[ 584.109556] (detected by 0, t=5252 jiffies, g=2953, q=74 ncpus=2) +[ 584.109561] Sending NMI from CPU 0 to CPUs 1: +[ 584.109587] NMI backtrace for cpu 1 skipped: idling at native_safe_halt+0xb/0x10 +[ 584.110564] rcu: rcu_preempt kthread timer wakeup didn't happen for 5251 jiffies! g2953 f0x0 RCU_GP_WAIT_FQS(5) ->state=0x402 +[ 584.110585] rcu: Possible timer handling issue on cpu=1 timer-softirq=8006 +[ 584.110597] rcu: rcu_preempt kthread starved for 5252 jiffies! g2953 f0x0 RCU_GP_WAIT_FQS(5) ->state=0x402 ->cpu=1 +[ 584.110614] rcu: Unless rcu_preempt kthread gets sufficient CPU time, OOM is now expected behavior. +[ 584.110645] rcu: RCU grace-period kthread stack dump: +[ 584.110658] task:rcu_preempt state:I stack:0 pid:15 ppid:2 flags:0x00004000 +[ 584.110667] Call Trace: +[ 584.110672] <TASK> +[ 584.110688] __schedule+0x351/0xa20 +[ 584.110699] ? rcu_gp_cleanup+0x480/0x480 +[ 584.110704] schedule+0x5d/0xe0 +[ 584.110705] schedule_timeout+0x94/0x150 +[ 584.110709] ? __bpf_trace_tick_stop+0x10/0x10 +[ 584.110714] rcu_gp_fqs_loop+0x141/0x4c0 +[ 584.110717] rcu_gp_kthread+0xd0/0x190 +[ 584.110720] kthread+0xe9/0x110 +[ 584.110725] ? kthread_complete_and_exit+0x20/0x20 +[ 584.110728] ret_from_fork+0x22/0x30 +[ 584.110735] </TASK> +[ 584.110736] rcu: Stack dump where RCU GP kthread last ran: +[ 584.110747] Sending NMI from CPU 0 to CPUs 1: +[ 584.110757] NMI backtrace for cpu 1 skipped: idling at native_safe_halt+0xb/0x10 + + ``` + +we can reproduce on our R630 cluster easily, but it is OK on R730 cluster and R740 cluster. +Steps to reproduce: +1. create and run an VM +2. migrate the vm to other host +3. it failed with message +Additional information: +i downgrade the pve-qemu-kvm from 8.1.2-4 to 8.0.2-3, same problem. |