Solaris 10 zones and capped memory

victor...@gmail.com

unread,

Jul 15, 2009, 2:01:57 PM7/15/09

to

What is the best way to analyze memory usage inside a zone with capped
memory? I have a server with numerous zones all with capped memory. If
I run prtconf inside the zone or top the zone reports total physical
memory in the zone instead of it's capped limit. Also I believe rcapd
in the global zone is working pretty hard; it's used about 305 hours
of cpu time in 68 days.

Thanks for any tips or insight.

Vic

unread,

Jul 15, 2009, 3:17:30 PM7/15/09

to

On Jul 15, 2:01 pm, "victor.en...@gmail.com" <victor.en...@gmail.com>
wrote:

Looking into rcapstat now...

kartikvashishta108

unread,

Jul 15, 2009, 7:30:25 PM7/15/09

to

prstat in the non global zone will show memory usage. zonecfg will
show the capped memory size. prstat with the -Z option will show about
zones.

# prstat -Z
PID USERNAME SIZE RSS STATE PRI NICE TIME CPU PROCESS/
NLWP
741 root 8704K 5744K sleep 59 0 10:55:27 6.0% svc.configd/
14
763 root 8448K 6560K wait 9 0 10:59:37 6.0% svc.configd/
14
12968 root 4128K 3504K cpu1 59 0 0:00:00 0.7% prstat/1
739 root 9280K 7344K wait 48 0 2:13:00 0.4% svc.startd/
16
761 root 9248K 7072K sleep 59 0 2:11:15 0.4% svc.startd/
18
12780 root 4088K 3472K sleep 59 0 0:00:00 0.1% prstat/1
1 root 2912K 1184K sleep 59 0 0:09:49 0.1% init/1
3021 noaccess 214M 134M sleep 59 0 0:10:44 0.1% java/17
13137 root 1664K 1264K wait 48 0 0:00:00 0.1% svc-snmpdx/
1
2998 noaccess 218M 138M sleep 59 0 0:10:34 0.1% java/17
2569 root 214M 136M sleep 59 0 0:09:37 0.1% java/17
683 root 25M 15M sleep 59 0 0:03:01 0.0% Xsun/1
942 root 6296K 4160K sleep 59 0 0:00:20 0.0% nscd/28
1020 root 6360K 4016K sleep 59 0 0:02:13 0.0% nscd/29
729 root 13M 2672K sleep 59 0 0:02:12 0.0% dtgreet/1
154 root 7000K 4496K sleep 59 0 0:00:25 0.0% nscd/31
1346 postgres 49M 38M sleep 59 0 0:01:42 0.0% postgres/1
1828 root 4152K 1816K sleep 59 0 0:01:54 0.0% syslogd/13
1745 root 4216K 2272K sleep 59 0 0:01:51 0.0% syslogd/13
907 root 6192K 3568K sleep 59 0 0:01:16 0.0% nscd/29
9899 unix 2704K 2024K sleep 59 0 0:00:00 0.0% telnet/1
1046 root 3104K 1712K sleep 59 0 0:00:51 0.0% xntpd/1
1022 root 3104K 1760K sleep 59 0 0:00:48 0.0% xntpd/1
1179 root 3064K 1576K sleep 59 0 0:00:51 0.0% xntpd/1
233 root 3104K 1448K sleep 100 - 0:00:42 0.0% xntpd/1
752 root 8496K 3568K sleep 59 0 0:00:36 0.0% svc.configd/
13
357 root 1664K 768K sleep 59 0 0:00:00 0.0% utmpd/1
512 root 4264K 1960K sleep 59 0 0:00:00 0.0% syslogd/13
333 root 7040K 4800K sleep 59 0 0:00:10 0.0% inetd/4
329 root 2512K 1080K sleep 59 0 0:00:00 0.0% sac/1
ZONEID NPROC SWAP RSS MEMORY TIME CPU ZONE
1 33 211M 259M 17% 13:25:21 6.6% ATPUGSAD
2 29 182M 238M 16% 13:26:56 6.5% NAUEEREEFS
0 52 94M 110M 7.1% 0:18:33 0.9% global
3 34 183M 244M 16% 0:12:06 0.2% AALOROTOM

Total: 148 processes, 575 lwps, load averages: 3.28, 3.27, 3.25

Kartik Vashishta

Alexander J. Maidak

unread,

Jul 19, 2009, 11:56:54 AM7/19/09

to

If you're on Solaris 10u7 or later the zone will report the correct value
for its total memory in top if you're using capped-memory physical value.

Using rcapstat -z 1 or something like that from the globalzone will give
the rcap activity. Also look at prstat -Z