We recently (about 6 weeks ago) finialized our RAC install. We are now in production mode.

At first I occassionally would see the olsnodes process come and go with minor spike levels (say 5% every 30 seconds). Lately it has been eating more and more CPU cycles and we are now at 25% constantly.

When I run olsnodes from the command prompt it takes about 8-10 seconds to come back - see below. It find the two nodes and then just hangs there from about 9 out of the 10 seconds.

[oracle@NS-I-ORA1 oracle]$ time /u01/app/oracle/oracle/product/10.2.0/crs/bin/olsnodes
ns-i-ora1
ns-i-ora2

real 0m10.145s
user 0m5.980s
sys 0m4.170s

Here's my system info

[oracle@NS-I-ORA1 oracle]$ uname -a
Linux NS-I-ORA1 2.4.21-40.ELsmp #1 SMP Thu Feb 2 22:22:39 EST 2006 i686 i686 i386 GNU/Linux

08:54:27 SQL> select * from v$version
08:54:33 2 /

BANNER

---

Oracle Database 10g Enterprise Edition Release 10.2.0.2.0 - Prod
PL/SQL Release 10.2.0.2.0 - Production
CORE 10.2.0.2.0 Production
TNS for Linux: Version 10.2.0.2.0 - Production
NLSRTL Version 10.2.0.2.0 - Production

Anyone run into the same thing? It seemed to have gradually worked it's way up. The second node seems to be doing OK without such dramatic usage. I notice that the pid is constantly changing.

Any help is appreciated...

JB

Hi.

Take a look at $ORA_CRS_HOME/log/<node>/client.

More you have .log files in this directory, slower is client-based commands such as olsnodes...

You may cron a batch script to clean up this log directory.

Edited by: ASoucisse on May 5, 2009 3:12 PM

When executing olsnodes successfully, it scans $CRS_HOME/log/<node name>/client directory and generates a new cssxxx.log file.

Oracle said:
Remove the log files under $ORA_CRS_HOME/log/<nodename>/client/css*.log in regular intervals with a cronjob.

Unpublished Bug 6004127 has been fixed in 11.1.0.7 and 10.2.0.4 CRS bundle Patch #2 onwards. Bundle Patch information is available