[OpenIndiana-discuss] System becoming unresponsive

[Matt Connolly]

Hi, my OI machine that I built recently has become unresponsive a number of times, requiring a hard reset.

Today this happened when I had an ssh connection open, running "top". This is the last screen I saw before all my ssh connections died and the machine stopped responding to all network activity.

load averages:  28.4,  11.5,  6.70;               up 3+00:50:14                                                           09:33:34
151 processes: 135 sleeping, 13 running, 3 on cpu
CPU states:  0.1% idle,  3.6% user, 96.3% kernel,  0.0% iowait,  0.0% swap
Kernel: 319 ctxsw, 15 trap, 517572 intr, 334874 syscall
Memory: 16G phys mem, 2054M free mem, 8054M total swap, 8054M free swap

   PID USERNAME NLWP PRI NICE  SIZE   RES STATE    TIME    CPU COMMAND
  1063 matt        8  59    0 1066M 1054M cpu/5  221:45 43.67% qemu-kvm-system
  4557 matt        7  59    0 2094M 2082M cpu/2    1:09  7.50% qemu-kvm-system
  1062 matt        8  59    0 1077M 1065M run    375:51  0.06% qemu-kvm-system
---snip---

(full page at https://gist.github.com/1597508 )

(The system is an Intel S1200-BTL motherboard with Xeon E3 processor and 16GB ECC ram.)


Once unresponsive, every now and then the HD activity led blinks.

Under normal circumstances, the load average is in the 1-4 range. The machine is still under fairly light use until I get some more confidence in it.

After giving the machine a hard reset, I don't see anything in the fault log or in /var/adm/messages indicating a failure or anything unusual. (kvm is a bit noisy, but it's like that all the time, so I'm not sure that that could be cause for lock up after 5 days).


Any ideas?

What could be causing so much cpu usage in the kernel?


Thanks,
Matt.