I'm not an expert, but I've been hunting down a killer performance problem
for a while now. It seems this may be the cause.
At peak load, our database slows to a trickle. The CPU and disk utilization
are normal - 20-30% used CPU and disk performance good.
All of our postgres processes end up
:00 0.70% 0.10% postgres
95959 pgsql 40 96048K 34268K sbwait 2 0:00 0.70% 0.10% postgres
-Original Message-
From: Tom Lane [mailto:[EMAIL PROTECTED]
Sent: Thursday, September 23, 2004 1:06 PM
To: Jason Coene
Cc: 'Mr Pink'; 'Scott Kirkwood'; [EMAIL PROTECTED]
Subject: Re
physical access to the machine to turn
HT off.
Thanks,
Jason
-Original Message-
From: Gaetano Mendola [mailto:[EMAIL PROTECTED]
Sent: Thursday, September 23, 2004 1:41 PM
To: Jason Coene
Subject: Re: Caching of Queries
Jason Coene wrote:
Hi Tom,
Easily recreated with Apache
until it had completed 800 requests. Unless I'm missing something, there's
more than the new connection IO load here.
Jason
-Original Message-
From: Jason Coene [mailto:[EMAIL PROTECTED]
Sent: Thursday, September 23, 2004 3:08 PM
To: [EMAIL PROTECTED]
Cc: [EMAIL PROTECTED]; [EMAIL
for while ab was running, from start to finish.
Any ideas?
Jason
-Original Message-
From: [EMAIL PROTECTED] [mailto:pgsql-performance-
[EMAIL PROTECTED] On Behalf Of Josh Berkus
Sent: Thursday, September 23, 2004 8:06 PM
To: Jason Coene; [EMAIL PROTECTED]
Cc: [EMAIL PROTECTED]; [EMAIL
My guess is that all the queries that involves the columns that are
being indexed need to
be rewritten to use the newly created indexes to avoid the performance
issues. The reason
is that REINDEX does not help either. Does it make sense?
Qing,
Generally, adding new indexes blindly will
Thanks for all the feedback. To clear it up, we are definitely not CPU
bound at the moment. Any slowdown seems to be disk dependant, or from to
serialization due to a long query (due to disk).
We do have a lot of INSERT/UPDATE calls, specifically on tables that track
user sessions, then of
Right. The point is: is your i/o bottle neck on the read side or the
write side. With 10-30 inserts/sec and fsync off, it's definitely on
the read side. What's interesting is that such a low insert load is
causing i/o storm problems. How does your app run with fsync on?
With
-Original Message-
From: Rod Taylor [mailto:[EMAIL PROTECTED]
Sent: Wednesday, August 11, 2004 5:46 PM
To: Jason Coene
Cc: 'Merlin Moncure'; Postgresql Performance
Subject: Re: [PERFORM] Hardware upgrade for a high-traffic database
I'm wondering why our PG server is using so
Hi Rod,
I was looking at top and vmstat - which always show under 300MB
Active.
We may hit 400MB at peak. Everything I see (though this isn't my area
of
expertise) points to most of the memory simply being unused. Results
below,
am I missing something?
This looks fine. The
You mean you are doing
SELECT ... WHERE userid = 42 ORDER BY timestamp DESC LIMIT 5;
and hoping that separate indexes on userid and timestamp will get the
job done? They won't. There are only two possible plans for this,
neither very good: select all of user 42's posts and sort them, or
this list, it seems to me that our settings are
reasonable given our usage, and that a disk upgrade is likely in order.
I'd love to hear any suggestions.
Thanks,
Jason
-Original Message-
From: Rod Taylor [mailto:[EMAIL PROTECTED]
Sent: Tuesday, August 10, 2004 7:07 PM
To: Jason Coene
Cc
a vmstat output, while running dd. The RAID array is tw0. It
does show the tw0 device getting significantly more work, numbers not seen
during normal operation.
Thanks,
Jason Coene
Gotfrag eSports
585-598-6621 Phone
585-598-6633 Fax
[EMAIL PROTECTED]
http://www.gotfrag.com
-Original Message
13 matches
Mail list logo