Forum OpenACS Development: Re: OpenACS performance issues

Collapse
8: Re: OpenACS performance issues (response to 1)
Posted by Dave Bauer on 03/05/08 06:25 PM
Yes, sorry I forgot to mention that! Your kernel max shared memory segment value needs to be increased to allow postgresql to use that much memory.
Collapse
9: Re: OpenACS performance issues (response to 8)
Posted by Matthew Coupe on 03/05/08 06:29 PM
What value would you recommend for that if I'm going for 128MB in the shared buffers?
Collapse
11: Re: OpenACS performance issues (response to 9)
Posted by Dave Bauer on 03/05/08 06:33 PM
This is the correct setting

# increase shared memory limit for postgres
kernel.shmmax = 134217728

for 128mb.

Collapse
17: Re: OpenACS performance issues (response to 11)
Posted by Matthew Coupe on 03/06/08 01:04 PM
I have kernel.shmmax set to 134217728 but then when I try to set shared_buffers to 128MB in the PostgreSQL config it won't start the postgresql service again?
Collapse
18: Re: OpenACS performance issues (response to 17)
Posted by Matthew Coupe on 03/06/08 01:09 PM
For information:

The database .dmp file is approximately 1GB. We run vacuum --analyze every hour and --full --analyze daily. With a DB that size do we need to increase memory by alot? I actually have the memory at hand to upgrade to 12GB but we haven't had chance to install it yet.

I've installed the latest dump file on a rubbish server and it is proportionally slow, so that should count out any other issues such as location or Red Hat OS issues.