Forum OpenACS Development: Re: OpenACS performance issues

28: Re: OpenACS performance issues (response to 1)

Posted by Gustaf Neumann on 03/07/08 09:52 AM

Good news! i guess, this means that you are happy with the performance now. maybe a conclusion we can draw is that if people realize a performance degration when upgrading to newer postgres versions (e.g. 8.2.5) is to check the shared buffers. Maybe postgres 8.2.5 suffers more than previous versions from insufficient shared buffers.

Note, that if you have now 96MB of shared buffers,
you are sill on the rather low side with your 4GB of memory.
Here are some old recommendations from postgres tuning:

* Start at 4MB (512) for a workstation
* Medium size data set and 256-512MB available RAM: 16-32MB (2048-4096)
* Large dataset and lots of available RAM (1-4GB): 64-256MB (8192-32768)

http://www.varlena.com/GeneralBits/Tidbits/perf.html#shbuf

Use "ipcs -m" and "ipcs -ml" on Linux to find memory usage and current limits.

29: Re: OpenACS performance issues (response to 28)

Posted by Matthew Coupe on 03/07/08 10:44 AM

Yes, I'm feeling a bit better about it now!

I'm going to play around with the values next week and try to optimise the performance even further. We're upping to 10GB of RAM in the server next week so I should be able to increase it even more.

SHMMAX was the only problem still outstanding. I'm going to try doubling that figure and then increase the Shared Buffers accordingly.

There are so many variables to consider when upgrading it's almost like starting all over again!

I'll try the ipcs-m and ipcs-ml commands when I get back to the office and report back. hopefully this thread will be useful for others wanting to optimise postgres performance.

Thank you so much for all of the replies to this post. Without you all putting in the time this project wouldn't be so successful.

Matthew