Re: Optimiser desicion bringing system to its knees?

From: Christopher Browne <cbbrowne(at)acm(dot)org>
To: pgsql-general(at)postgresql(dot)org
Subject: Re: Optimiser desicion bringing system to its knees?
Date: 2003-11-07 03:48:31
Message-ID: m38yms4z5c.fsf@wolfe.cbbrowne.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general

In the last exciting episode, craig(at)ucw(dot)com(dot)au ("Craig O'Shannessy") wrote:
> I'm not onsite, but I just heard back from a colleage that rushed in to
> try exactly that. It does indeed seem to be fixed in 7.4RC1. I assume
> this will be pretty stable, and there seems to be heaps of enhancements,
> so I think I'll just skip 7.3.x. I'm not doing anything funky with
> postgreSQL, but if someone has a good reason for me NOT putting 7.4RC1
> onto a production system, please let me know ;)

Would you rush FreeBSD 5.1 into production when they're still
experimenting with it?

Would you rush Linux 2.6 into production when they still haven't got
the .0 release of that?

I'm doing some active work testing 7.4RC1 with some apps on the basis
that by the time the software is ready to deploy, PG will surely have
reached version 7.4.0 or 7.4.1. But if I planned something for
production TOMORROW (e.g. - as in Friday, November 7, the day after
today, November 6th), I would be looking at 7.3.4.

Don't misinterpret that as any sort of "vote of nonconfidence;" it's
just the notion of having a modicum of conservatism, and staying a
_small_ step away from the bleeding edge of the abyss, at least when
looking at production systems.

I similarly haven't yet had opportunity to test out Jan Wieck's ARC
patch, improving the way the system copes with VACUUM when under heavy
load. Of vast ultimate interest though it may be, I simply haven't
got a test set ready to _properly_ torture the patch, so it makes
sense to wait and maybe look at it a little later as it matures.

There's an old line about rushing in where angels fear to tread.
Putting not-yet-released software into production is one of those
places...
--
(format nil "~S(at)~S" "cbbrowne" "ntlug.org")
http://www.ntlug.org/~cbbrowne/
Rules of the Evil Overlord #104. "My undercover agents will not have
tattoos identifying them as members of my organization, nor will they
be required to wear military boots or adhere to any other dress
codes." <http://www.eviloverlord.com/>

In response to

Browse pgsql-general by date

  From Date Subject
Next Message Bruce Momjian 2003-11-07 04:02:01 Re: Cartesian product bug?
Previous Message sgupta5 2003-11-07 03:47:20