There is no secret that there are memory leaks in our CIM Server and it seems that not only XenServer has such problems. Recetly I spent some time helping my colleague troubleshooting problem like this and decided to share following the procedures
Procedure 1 :. Using database files
- Set destination and the core file pattern:
sysctl -w kernel.core_pattern = "... / var / tmp / core% e% h% p "-q kernel sysctl .core_pattern
above the set file core sales pitch commands is as follows:
/var/tmp/core.
[ . - Change
/etc/init.d/openpegasus
file and addulimit -c unlimited
line before cimserver begins.
Or apply patch below on/etc/init.d/openpegasus
*** / etc /init.d/openpegasus .00 04/05/2013 14: 39: 48.000000000 0530 --- /etc/init.d/openpegasus 05/04/2013 14: 31: 42.000000000 0530 ******* **** **** start () {*** **** 11.16 --- 11.17 ---- stop () {echo -n "stop OpenPegasus ..." ulimit -c unlimited + / opt / OpenPegasus / bin / cimserver -s echo "done." }
To apply this patch:
- save a text file in XenServer dom0, namely:
/root/cim.patch
- run
Patch
- save a text file in XenServer dom0, namely:
- cimserver Restart
Service OpenPegasus restart
- When cimserver begins consume abnormal amount of memory kill cimserver with SIGABRT (6) signals.
-SIGABRT kill $ (pgrep cimserver)
- Check new file / var / tmp / core.
Procedure 2 :. Using part of valgrind
- Stop cimserver watchdog
touch / opt / xs-cem-cmpi / watchdog-off
- Set cimserver running in the foreground:
/ opt / OpenPegasus / bin / daemon cimconfig -s = false -p
- Start cimserver using valgrind
valgrind --leak-check = - -log complete file = / root / cimserver.valgrind% p.Log --trace-children = yes / opt / OpenPegasus / bin / cimserver
- When problem occurs stop valgrind and collect log file / var / tmp
Follow remaining steps back cimconfig original configuration. - cimserver Set to run in the background:
/ opt / OpenPegasus / bin / cimconfig -s daemon = true -p
- Enable monitoring:
rm - vF / opt / xs-cem-IPMC / watchdog-off
0 Komentar