Hi all,
My problem has not yet been resolved. I had to reload my machine because
of routine maintenance, so I have to wait for another 60+ days of uptime
for it to re-appear. Thanx for all the people that answered and tried to
help.
Although Ed Finch did not answer my question directly he send me this
info:
#
# FWIW, I've written a tool to graph the output from sar; see
# http://www.vais.net/~efinch for more information.
# Best regards,
# Ed Finch
I looked at it and love it. I think Ed must have burned a lot of
midnight oil on this. It is really something worth to look into.
u-kevin@megami.veritas.com's reply confirms to me that somebody else
also saw this problem. His answer :
# I have always been suspicious of that field's contents, but have not
# tracked down why it doesn't seem to reflect reality.
# vmstat includes wio in idle time. I generally believe sar or iostat
-c
# on that one. iostat -xcn is best for CPU/IO, and vmstat for the
paging
# stuff.
# You might want to use top instead. gives pretty comprehensive output
# and generally gives you a good idea as to the culprit.
# l & h,
# kev
Casper Dik answered:
# When a processor is in wio it's really idle and available
# for other tasks, but there are outstanding read requests.
# There's not much conflict when you look at it that way.
# Casper
If Casper's comment is true it confuses me even more because I have
monitored this behavior for +- 3 weeks and tried to find a reason for it
before I mailed the group. A outstanding read request for that long I
don't think so because if you look at my original question added below
this machine with its 14CPU's and 4GB memory is relatively in an idle
state.
Thanx again for everybody and a great sun-managers list. It is great to
have a lot of friends just a keyboard away when you are in a spin. I
will add another summary if I find something worth to share.
Regards
Gert Marais
My original question :
Hi Sun-managers,
I have picked up a strange problem on my E6000 regarding the different
methods of performance stats you can collect. The machine is running
some decision support stuff from front-end clients. Included is
different stats I have collected on the machine. Note the vmstat output
regarding the runnable processes which is swapped or sleeping. You can
also see that this 4Gb memory is not swapping. Sar gives very high wio
and low CPU idle where vmstat gives very high idle time and iostat gives
0 waiting %. Nobody is complaining about any lack of performance. I
would appreciate if you can shed some light on this strange output.
Sorry for the size of this mail item, I have tried to keep only the
important stuff.
#
#sar -u 5 5
#
SunOS SAPS2 5.6 Generic_105181-06 sun4u 09/17/98
11:42:37 %usr %sys %wio %idle
11:42:42 9 1 44 46
11:42:47 9 2 49 41
11:42:53 9 1 48 42
11:42:58 8 1 41 50
11:43:03 9 1 36 54
Average 9 1 44 47
#
#vmstat 5
#
procs memory page disk faults
cpu
r b w swap free re mf pi po fr de sr s0 s1 s2 s2 in sy cs us
sy id
0 0 36 2610376 63408 7 0 803 241 929 0 90 6 1 1 2 580 1538 814 8
2 90
0 0 36 2610376 63424 9 29 611 297 660 0 47 2 1 0 0 463 1149 760 8
1 91
0 0 36 2610376 63344 6 0 873 438 964 0 68 0 1 0 0 582 1267 915 9
1 90
0 1 36 2610376 63488 8 0 712 441 804 0 46 0 0 0 0 559 1175 857 8
1 90
0 0 36 2610368 63480 3 0 564 235 593 0 48 1 0 1 0 378 1739 535 8
1 91
#
#vmstat -S 5
#
procs memory page disk faults
cpu
r b w swap free si so pi po fr de sr s0 s1 s2 s2 in sy cs us
sy id
0 1 36 2610368 63456 0 0 516 427 710 0 38 3 7 4 2 500 6823 857 9
1 90
0 1 36 2610368 63368 0 0 948 379 878 0 77 7 5 2 1 624 1125 936 9
1 90
0 0 36 2610368 63472 0 0 505 254 603 0 44 0 0 2 1 443 2432 674 8
1 91
0 0 36 2610368 63472 0 0 459 388 633 0 31 0 1 0 0 536 1602 792 8
1 91
0 1 36 2610368 63440 0 0 1072 451 1056 0 79 0 0 0 0 632 1097 981 9
1 90
#
#mpstat 5
#
CPU minf mjf xcal intr ithr csw icsw migr smtx srw syscl usr sys wt
idl
0 7 2 285 7 0 145 9 6 4 0 414 6 3 61
31
1 7 2 310 13 5 155 9 6 4 0 455 7 3 60
30
4 7 2 260 7 0 147 9 6 4 0 413 6 3 61
31
5 8 2 307 7 0 162 10 6 4 0 485 7 3 60
30
8 7 2 251 7 0 147 9 6 4 0 411 6 2 61
31
9 8 2 100 116 110 141 8 6 5 0 440 6 3 60
31
12 7 2 69 235 29 144 9 6 4 0 395 6 2 61
31
13 8 2 92 398 386 145 9 6 8 0 474 7 3 60
31
14 8 3 290 35 26 167 10 7 4 0 524 8 3 59
30
16 7 2 232 8 1 146 9 6 4 0 425 6 2 61
31
17 7 2 264 31 23 158 9 6 4 0 482 7 3 60
30
20 8 2 322 8 0 165 10 7 4 0 526 8 3 60
30
24 7 2 294 22 14 158 10 7 4 0 494 7 3 60
30
28 7 2 296 10 2 157 10 7 4 0 474 7 3 60
30
#
#iostat
#
device r/s w/s kr/s kw/s wait actv svc_t %w %b
c1t0d0 1.7 0.0 9.3 0.0 0.0 0.0 8.2 0 1
c3t5d0 0.3 0.3 2.7 2.7 0.0 0.0 5.7 0 0
c4t4d2 0.0 0.3 0.0 2.7 0.0 0.0 7.1 0 0
c4t4d3 0.0 7.7 0.0 61.2 0.0 0.1 8.6 0 7
c5t4d1 0.0 0.7 0.0 0.0 0.0 0.0 0.0 0 0
c5t4d2 4.3 10.3 220.9 82.5 0.0 0.1 6.9 0 7
c6t5d0 0.0 0.3 0.0 2.7 0.0 0.0 6.2 0 0
c2t0d1 2.7 4.7 50.2 33.4 0.0 0.1 9.2 0 5
c2t0d2 0.7 2.3 17.1 17.1 0.0 0.0 6.3 0 2
c2t0d3 0.0 1.3 0.0 7.8 0.0 0.0 2.6 0 0
c2t0d4 3.7 3.3 91.2 26.6 0.0 0.1 13.1 0 8
c2t0d5 2.0 0.7 41.9 5.3 0.0 0.0 16.1 0 4
c2t0d6 0.0 1.0 0.0 6.0 0.0 0.0 2.7 0 0
c2t1d1 0.7 5.0 28.9 32.4 0.0 0.0 4.4 0 2
c2t1d2 0.7 1.3 22.5 9.5 0.0 0.0 12.5 0 2
c2t1d3 0.0 2.7 0.0 18.8 0.0 0.0 2.0 0 1
c2t1d4 2.0 2.0 57.9 13.3 0.0 0.1 15.9 0 5
c2t1d5 3.3 1.0 74.9 6.3 0.0 0.1 13.4 0 4
c2t1d6 1.0 0.7 22.3 5.3 0.0 0.0 14.0 0 1
c2t2d1 2.3 3.0 48.2 18.3 0.0 0.1 16.1 0 5
c2t2d2 0.3 2.7 0.3 21.3 0.0 0.0 2.7 0 1
c2t2d3 0.0 3.0 0.0 18.8 0.0 0.0 7.6 0 2
c2t2d4 2.7 5.3 25.0 39.9 0.0 0.1 7.0 0 6
c2t2d5 2.3 2.0 35.6 15.0 0.0 0.0 10.4 0 4
c2t2d6 1.0 3.3 23.0 26.6 0.0 0.0 5.5 0 2
c2t3d1 2.0 3.7 48.2 25.0 0.0 0.0 8.1 0 4
c2t3d2 0.7 0.7 5.3 5.3 0.0 0.0 10.6 0 1
c2t3d3 0.7 2.0 10.6 13.1 0.0 0.0 6.5 0 2
c2t3d4 3.0 0.7 49.6 5.3 0.0 0.1 15.7 0 5
c2t3d5 0.3 1.3 2.7 10.6 0.0 0.0 6.5 0 1
c2t3d6 1.0 2.0 15.3 15.3 0.0 0.0 7.3 0 2
TOTAL w/s 75.3 548.1
r/s 39.4 903.6
#
# uptime
#
12:01pm up 38 day(3), 15 hr(s), 3 users, load average: 0.46, 0.85, 1.38
#
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:12:50 CDT