Re: Indexes slower when used in decending vs. ascending order?

I'm basically stuck with that as well: we run a custom kernel and a
stack of drivers that have no support past 7.2 (at least presently) so
it looks like I'll just need to work out a way of getting 8.0 to compile
on rh7.2 (or patching postgres 7.3 with the bugfix and recompiling :)

None of those look like fun, so this issue will probably be added to our
stack of "reasons to upgrade to FC5"

:)

- alasdair

-----Original Message-----
From: Steve Crawford [mailto:scrawford@pinpointresearch.com]
Sent: Wednesday, April 12, 2006 9:43 AM
To: Tom Lane
Cc: Alasdair Young; pgsql-novice@postgresql.org
Subject: Re: [NOVICE] Indexes slower when used in decending vs.
ascending order?

> Um ... I work for Red Hat, so I think I have to slip you the bad news
> twice today: no sane person would still be running that OS version
> either.  7.2 was obsolete about the time I went to work at Red Hat (in

> 2001).

Sadly, some users are stuck. For instance if you have Dialogic D240SC/T1
Rev 1 cards then the latest drivers you can get are the 5.x and they are
very picky about kernel versions, streams libs, etc.

If you want to build a supported system, you must start with Red Hat
7.2/7.3 and must, among other detailed steps, apply specific kernel
patches. Or you can go through a similar set of steps on Windows. I'll
take Linux - even an old version.

Fortunately I have it on authority of an Intel employee that they will
be open-sourcing the Dialogic drivers sometime in the next year so over
time this problem will probably disappear.

Cheers,
Steve