Hi Gino,
Could you login as root, and run
# mdb -k
At mdb prompt, type
> ::cpuinfo
One of the CPUs is probably running sched. Use the value under heading
"THREAD" for the CPU, and do
> put-your-thread-value-here::findstack -v
For example,
> fffffffed378a760::findstack -v
Please let me know what the stack trace says. Could you also send me
the output of "psrinfo -vp"?
The symptom of the problem you are seeing is similar to that with
6563696 cpu idle loop takes up 100% system time on a freshly booted
wolf blade
But I wouldn't know for sure until I have more information.
Sherry
On Thu, Jul 12, 2007 at 08:18:16AM -0700, Gino wrote:
> Hi All,
>
> snv67, 4xOpteron.
> We have some performance problems on this server.
> In particular mpstat is reporting CPU#1 at 93% sys.
> Is there a way to find out what that cpu is doing?
>
> svr11@/test# lockstat -kIW -D 20 sleep 30
>
> Profiling interrupt: 11648 events in 30.026 seconds (388 events/sec)
>
> Count indv cuml rcnt nsec Hottest CPU+PIL Caller
> -------------------------------------------------------------------------------
> 5710 49% 49% 0.00 855 cpu[0] mach_cpu_idle
> 1614 14% 63% 0.00 4278 cpu[3] (usermode)
> 862 7% 70% 0.00 1072 cpu[1] do_splx
> 858 7% 78% 0.00 1126 cpu[1]+11 splr
> 267 2% 80% 0.00 767 cpu[1] disp_getwork
> 204 2% 82% 0.00 3446 cpu[3] vmu_calculate_seg
> 195 2% 83% 0.00 3331 cpu[3] page_exists
> 192 2% 85% 0.00 2967 cpu[3] i_mod_hash_find_nosync
> 157 1% 86% 0.00 768 cpu[1] swtch
> 147 1% 88% 0.00 754 cpu[1] new_cpu_mstate
> 135 1% 89% 0.00 767 cpu[1] disp
> 88 1% 90% 0.00 754 cpu[1] tsc_gethrtimeunscaled
> 70 1% 90% 0.00 4373 cpu[3] lzjb_compress
> 67 1% 91% 0.00 4059 cpu[3] mutex_enter
> 64 1% 91% 0.00 797 cpu[1]+11 lock_set_spl
> 60 1% 92% 0.00 820 cpu[1]+11 disp_lock_exit
> 49 0% 92% 0.00 742 cpu[1] idle_enter
> 35 0% 92% 0.00 733 cpu[1] idle_exit
> 32 0% 93% 0.00 989 cpu[1] disp_lock_enter
> 26 0% 93% 0.00 4419 cpu[3] fsflush_do_pages
> -------------------------------------------------------------------------------
> svr11@/test# mpstat 1
> CPU minf mjf xcal intr ithr csw icsw migr smtx srw syscl usr sys wt idl
> 0 0 0 0 2224 1220 0 0 0 2 0 0 0 1 0 99
> 1 13 0 0 283 278 21 0 0 2 0 59 0 93 0 7
> 2 0 0 0 84 72 15 0 0 2 0 13 0 0 0 100
> 3 1389 0 10 872 4 1201 312 0 12 0 6658 22 77 0 1
> CPU minf mjf xcal intr ithr csw icsw migr smtx srw syscl usr sys wt idl
> 0 0 0 0 2455 1375 0 0 0 13 0 0 0 1 0 99
> 1 3064 0 221 141 122 334 18 13 7 0 5614 11 84 0 5
> 2 0 0 0 286 198 14 0 0 0 0 5 0 0 0 100
> 3 102 0 18 627 2 1579 541 0 62 0 279482 28 66 0 7
> CPU minf mjf xcal intr ithr csw icsw migr smtx srw syscl usr sys wt idl
> 0 0 0 0 2349 1305 0 0 0 7 0 0 0 1 0 99
> 1 28 0 9 46 3 108 0 0 0 0 487 0 93 0 7
> 2 297 0 27 288 235 226 2 1 2 0 850 1 1 0 98
> 3 9107 0 99 270 4 996 280 0 17 0 108063 61 38 0 1
>
> tnx,
> gino
>
>
> This message posted from opensolaris.org
> _______________________________________________
> opensolaris-discuss mailing list
> opensolaris-discuss@opensolaris.org
--
Sherry Moore, Solaris Kernel Development http://blogs.sun.com/sherrym
_______________________________________________
opensolaris-discuss mailing list
opensolaris-discuss@opensolaris.org
