Load average in Veritas cluster

[Britto Sidhan]

Hi Gurus,

I've two nodes in cluster and the load average in active node is very high. Pl. find the uptime of active machine as mentioned below:

xyz@node-2 # uptime
3:39pm up 2 day(s), 3:32, 2 users, load average: 8.55, 9.03, 8.91

Pl. find the uptime of passive machine as mentioned below:

xyz@node-1 # uptime
3:40pm up 2 day(s), 3:34, 2 users, load average: 0.00, 0.00, 0.00

Kindly inform whether it's an application specific problem or a system related problem. Oracle database is running in Veritas cluster.

Attachment: ps-efl output from active cluster node.
Thanking you in advvance

Thanks and Regards
Britto Sidhan

[John Waller]

Looking at the output from the ps -ef the following line may give a clue to the high load average:
0 R root 9908 1 1 73 27 ? 46778 May 07 ? 772:29 /opt/billing/j2sdk1.4.2_05/bin/java

A CPU time of 772.29 seems a bit high so I believe it may be an application configuration issue.

[Steven E. Protter]

shalom Britto,

load is not a good means of knowing whether a system is overloaded. If there are respone problems and user apps run slowly, the load factor can still be very low and the problem severe.

The most common cause of actual problems is poorly written application code. Next most common is poorly written sql/plsql code to the database that will show up as i/o problems with little evidence in the load numbers.

SEP

[reju rajeev]

Mr Britto,

Poorly written application code is the issue . I hope fine tuning the Pl/Sql codes will solve your problem. Please ask your DBA to it.

Regards

Reju

[Britto Sidhan]

Thanks for all your kind help. I've informed the concerned person about this.


Thanking you all once again

Britto Sidhan