From: | "Merlin Moncure" <merlin(dot)moncure(at)rcsonline(dot)com> |
---|---|
To: | "Kari Lavikka" <tuner(at)bdb(dot)fi> |
Cc: | <pgsql-performance(at)postgresql(dot)org> |
Subject: | Re: Finding bottleneck |
Date: | 2005-07-28 13:02:18 |
Message-ID: | 6EE64EF3AB31D5448D0007DD34EEB3417DCF28@Herge.rcsinc.local |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-performance |
Kari Lavikka wrote:
> shared_buffers = 15000
you can play around with this one but in my experience it doesn't make
much difference anymore (it used to).
> work_mem = 1536 # min 64, size in KB
this seems low. are you sure you are not getting sorts swapped to disk?
> fsync = true # turns forced synchronization on or
off
does turning this to off make a difference? This would help narrow down
where the problem is.
> commit_delay = 80000 # range 0-100000, in microseconds
hm! how did you arrive at this number? try setting to zero and
comparing.
> stats_start_collector = true
> stats_command_string = true
with a high query load you may want to consider turning this off. On
win32, I've had some problem with stat's collector under high load
conditions. Not un unix, but it's something to look at. Just turn off
stats for a while and see if it helps.
good luck! your hardware should be more than adequate.
Merlin
From | Date | Subject | |
---|---|---|---|
Next Message | Chris Hoover | 2005-07-28 14:38:06 | Fwd: Help with view performance problem |
Previous Message | Merlin Moncure | 2005-07-28 12:20:35 | Re: Performance problems testing with Spamassassin 3.1.0 Bayes module. |