From: | Curt Sampson <cjs(at)cynic(dot)net> |
---|---|
To: | Markus Wollny <Markus(dot)Wollny(at)computec(dot)de> |
Cc: | pgsql-general(at)postgresql(dot)org |
Subject: | Re: Urgent: Tuning strategies? |
Date: | 2002-06-26 05:03:32 |
Message-ID: | Pine.NEB.4.43.0206261358150.670-100000@angelic.cynic.net |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
On Tue, 25 Jun 2002, Markus Wollny wrote:
> There's an UPS connected to the server,
Which can also fail, and which does not protect against, e.g., a
kernel panic.
> backups are performed each day,
Well, even given the data loss, how long is it going to take you
to do a restore from those backups, compared to just letting the
database clean things up and continue on? Or do you not care about
downtime, either?
> I don't see much point in not setting fsync to false for faster updates.
> And community-data's not that valuable to us, we can without doubt risk
> loosing one day's updates.
fsync or not makes little speed difference if your log has its own disk.
> Ah, that might be one big issue - I often noticed several backends
> growing far beyond 200MB size, which got me worried...
Right. Also, for your queries doing sorts, check out the query stats
("set show_query_stats yes"; results go to the error log) and see if
they're swapping. If they are, that's killing your performance.
cjs
--
Curt Sampson <cjs(at)cynic(dot)net> +81 90 7737 2974 http://www.netbsd.org
Don't you know, in this new Dark Age, we're all light. --XTC
From | Date | Subject | |
---|---|---|---|
Next Message | Carl Meyer | 2002-06-26 05:43:19 | unlimited undo/journaling |
Previous Message | Bruce Momjian | 2002-06-26 03:58:35 | Re: What is a tuple? |