From: John A Meinel <john@arbash-meinel.com>
To: Alvaro Herrera <alvherre@alvh.no-ip.org>
Cc: Josh Berkus <josh@agliodbs.com>
Cc: pgsql-performance@postgresql.org, "Jeffrey W. Baker" <jwbaker@acm.org>
Cc: Steve Poe <spoe@sfnet.cc>
Cc: paul@oxton.com
Subject: Re: PG8 Tuning
Date: Tue, 16 Aug 2005 11:33:43 -0500
Message-ID: <43021567.1000101@arbash-meinel.com> (raw)
In-Reply-To: <20050816162531.GB32695@alvh.no-ip.org>
References: <1536.217.45.209.171.1123763001.squirrel@www.gradwell.com>
<1123836428.19976.26.camel@amd64-laptop-spoe>
<1123779524.6664.1.camel@toonses.gghcwest.com>
<200508160912.32263.josh@agliodbs.com>
<20050816162531.GB32695@alvh.no-ip.org>
Alvaro Herrera wrote:
> On Tue, Aug 16, 2005 at 09:12:31AM -0700, Josh Berkus wrote:
>
>
>>However, you are absolutely correct in that it's *relative* advice, not
>>absolute advice. If, for example, you're using a $100,000 EMC SAN as your
>>storage you'll probably be better off giving it everything and letting its
>>controller and cache handle disk allocation etc. On the other hand, if
>>you're dealing with the 5 drives in a single Dell 6650, I've yet to encounter
>>a case where a separate xlog disk did not benefit an OLTP application.
>
>
> I've been asked this a couple of times and I don't know the answer: what
> happens if you give XLog a single drive (unmirrored single spindle), and
> that drive dies? So the question really is, should you be giving two
> disks to XLog?
>
I can propose a simple test. Create a test database. Run postgres,
insert a bunch of stuff. Stop postgres. Delete everything in the pg_xlog
directory. Start postgres again, what does it do?
I suppose to simulate more of a failure mode, you could kill -9 the
postmaster (and all children processes) perhaps during an insert, and
then delete pg_xlog.
But I would like to hear from the postgres folks what they *expect*
would happen if you ever lost pg_xlog.
What about something like keeping pg_xlog on a ramdisk, and then
rsyncing it to a hard-disk every 5 minutes. If you die in the middle,
does it just restore back to the 5-minutes ago point, or does it get
more thoroughly messed up?
For some people, a 5-minute old restore would be okay, as long as you
still have transaction safety, so that you can figure out what needs to
be restored.
John
=:->
Attachments:
signature.asc (application/pgp-signature, 252B) download
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients using the --to and --cc options:
reply via email
To: pgsql-performance@postgresql.org
Cc: john@arbash-meinel.com, alvherre@alvh.no-ip.org, josh@agliodbs.com, jwbaker@acm.org, spoe@sfnet.cc, paul@oxton.com
Subject: Re: PG8 Tuning
In-Reply-To: <43021567.1000101@arbash-meinel.com>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox