This makes sense, however, I am still a bit confused. You are stating that in
Solaris 9 microstate accounting gets only updated when the lwp is transitions
from one state to another. So, if the process was idle for a long time (idle
since before microstate accounting was started by prstat -m ??) then I can see
that idle state counters might be out of whack. However, I would not expect
this to happen if the process had some activity since the prstat -m was
initiated. I am probably missing something.. Here is prstat-m output piped
throgh
gcars0d7% cat proc_prstat-m
#/bin/sh
nawk '
BEGIN {
} { t=$3+$4+$5+$6+$7+$8+$9+$10
if ((t < 90)||(t>110)) print $0, t
}
END{
}'
gcars0d7%
Look at the PID 1357 - it is always the same thread, esd/1 and it certainly had
some activity since prstat -m was initiated (USE/SYS/VCX/ICX are non-0).
If this is a known issue, does it impact idle/SLP counts only? Can I assume
that USR/SYS are correct? (in that case I can derive the idle from the rest)
gcars0d7% prstat -mL 10 | proc_prstat-m
PID USERNAME USR SYS TRP TFL DFL LCK SLP LAT VCX ICX SCL SIG PROCESS/LWPID
** 1357 root 2.0 0.1 - - - - 69 - 330 27 1K 0 esd/1 71.1
18022 cemsbin 1.5 0.4 0.0 0.0 0.0 0.5 78 0.0 10 23 12 0 java/2 80.4
10405 root 1.4 0.0 - - - - 67 - 4 17 35 0 nco_objserv/22
8.4
18200 cemsbin 1.1 0.0 0.0 0.0 0.0 0.5 78 0.0 10 13 13 0 java/6 79.6
432 nortel 1.0 0.0 - - - - 67 - 10 13 11 0 java/6 68
18022 cemsbin 0.7 0.2 0.0 0.0 0.0 0.0 79 0.0 20 14 307 0 java/123 79.9
23215 root 0.9 0.0 - - - - 67 - 10 11 11 0 java/6 67.9
Total: 442 processes, 2083 lwps, load averages: 1.12, 1.52, 1.68 2087.32
PID USERNAME USR SYS TRP TFL DFL LCK SLP LAT VCX ICX SCL SIG PROCESS/LWPID
1 root 0.4 1.4 - - - - 0.0 - 717 27 21K 986 init/1 1.8
** 1357 root 1.4 0.1 - - - - 0.0 - 319 13 865 0 esd/1 1.5
383 root 0.2 0.1 - - - - 0.0 - 1K 0 4K 0 nscd/51 0.3
Total: 444 processes, 2077 lwps, load averages: 1.68, 1.62, 1.71 2082.01
PID USERNAME USR SYS TRP TFL DFL LCK SLP LAT VCX ICX SCL SIG PROCESS/LWPID
12551 nortel 0.0 33 - - - - 99 - 331 0 18K 0 cut/1 132
5029 nortel 2.6 8.6 - - - - 23 - 1K 112 16K 0 wam_zapOneGr/1
4.2
5004 nortel 2.0 6.9 - - - - 26 - 1K 93 10K 0 wam_zapOneGr/1
4.9
23215 root 5.5 0.0 - - - - 0.0 - 11 62 19 0 java/6 5.5
1 root 1.3 3.0 - - - - 0.0 - 1K 50 31K 1K init/1 4.3
** 1357 root 1.8 0.1 - - - - 0.0 - 324 18 1K 0 esd/1 1.9
25290 nortel 1.5 0.0 - - - - 0.0 - 11 29 15 0 java/6 1.5
383 root 0.1 0.3 - - - - 0.0 - 1K 4 5K 0 nscd/51 0.4
Total: 443 processes, 2099 lwps, load averages: 2.10, 1.71, 1.74 2104.55
PID USERNAME USR SYS TRP TFL DFL LCK SLP LAT VCX ICX SCL SIG PROCESS/LWPID
25290 nortel 3.0 0.0 - - - - 0.0 - 10 39 13 0 java/6 3
** 1357 root 2.1 0.0 - - - - 0.0 - 430 27 649 0 esd/1 2.1
12172 root 1.4 0.0 - - - - 0.0 - 11 37 12 0 java/2 1.4
1 root 0.4 0.4 - - - - 0.0 - 223 8 6K 307 init/1 0.8
10405 root 0.2 0.0 - - - - 0.0 - 10 7 191 0 nco_objserv/753
0.2
29183 root 0.1 0.1 - - - - 0.0 - 19 2 229 12 nco_pad/1 0.2
10405 root 0.1 0.0 - - - - 0.0 - 5 1 112 0 nco_objserv/754
0.1
432 nortel 0.1 0.0 - - - - 0.0 - 1K 1 594 0 java/41 0.1
Total: 437 processes, 2071 lwps, load averages: 2.09, 1.73, 1.75 2076.57
gcars0d7%
This message posted from opensolaris.org
_______________________________________________
perf-discuss mailing list
perf-discuss@opensolaris.org
