Showing results for 
Search instead for 
Do you mean 

Trucluster slowing down

Occasional Contributor

Trucluster slowing down

We have a performance problem with a 2 node Trucluster.
Following the boot is everything O.K.
After 4-5 week of uptime, the response time of shell commands and scripts
is getting to long. Specially the execution time of "ps ax" command is very long.
This results in timing out of scripts which are checking the cluster services.
There are 8 Oracle 8.1.7 instances on cluster, 2 big instances on DELTA, 6 smaller on GAMA.
The impact on Oracle is not so much as on shell commands and scripts.

OS is Tru64 V5.1A PK4.

The hw configuration is:
Current configuration:

DELTA:

HW: GS160 with 4 QBB only 3 populated
10 CPU (4/4/2/0) (QBB1/QBB2/QBB3/QBB4)
24 GB RAM (8/8/8/0)
3 I/O Riser (1/1/1/0)
3 I/O Drawers (1/1/1/0)
3 FC (2 Gbit ) adapters (1/1/1/0)
1 Gbit Eth (1/0/0/0)
2 MC (1/1/0/0)

1 hw partition with 3 RADs

GAMA:

HW: GS140
6 CPU
12GB RAM ( 4+4+4)
1 KFTHA
2 DWLPx
2 FC (1 Gbit) adapters (1/1)
1 Gbit Eth adapter (1/0)
2 MC adapters (1/1)

Any help appreciated!
2 REPLIES
Advisor

Re: Trucluster slowing down

Hi, i think is better you upgrade at PK6...
Regards Norman
Valued Contributor

Re: Trucluster slowing down

Hi,

If you aren't running collect(8) to gather historical and daily performance statistices, I'd recommend that you do that. If you do have collect data, it would be useful to see the graphs over time to see if memory is getting used up, or CPU utilization gradually increases or something along those lines. They key is that if there are trends (as you suggest) we would be able to see those trends with the collect data fairly quickly.

For more on collect, including how to set it up to run automatically see:

http://h30097.www3.hp.com/collect/collect_details.html

Greg
//Add this to "OnDomLoad" event