Re: [GENERAL] Testers needed ... - Mailing list pgsql-hackers

From Marc G. Fournier
Subject Re: [GENERAL] Testers needed ...
Date
Msg-id 20020416113728.G99298-100000@mail1.hub.org
Whole thread Raw
In response to Re: [GENERAL] Testers needed ...  (Tom Lane <tgl@sss.pgh.pa.us>)
Responses Re: [GENERAL] Testers needed ...  (Jan Wieck <janwieck@yahoo.com>)
List pgsql-hackers
On Tue, 16 Apr 2002, Tom Lane wrote:

> "Marc G. Fournier" <scrappy@hub.org> writes:
> > Could some ppl test out archives.postgresql.org and let me know if they
> > notice any differences in speed?
>
> Yup.  It's usable again!  What did you do?

Got more RAM installed :)  The archives have a buffer cache right now of
1.5Gig, and will be jumped to a full 3Gig as soon as the "special RAM"
gets in ... Rackspace had to special order in some RAM to bring it from
3Gig->4gig ... I'm suspecting they needed to switch to low-voltage RAM for
that, which they didn't have in stock ...

I've also moved the indices to one file system, while leaving the tables
themselves on the other (this box is limited to 2 hard drives,
unfortunately), which seems to be doign a better job of keeping disk I/O
down a bit ...

Am currently rebuilding the indices, so *everything* isn't in there yet,
but we're up to:

          Database statistics

    Status    Expired      Total
   -----------------------------
         0      61548      61734 Not indexed yet
       200          0      66252 OK
       404          0          8 Not found
   -----------------------------
     Total      61548     127994

and climbing fast ...

Oh, and, of course, its running on v7.2.1 now, which means that VACUUMs no
longer lock up the search ... last night, I had something like 367
indexers pounding away at the database *and* a VACUUM running *and* did a
search in <2 minutes (considering that is 367 simultaneous and active
connections to the database, I'm impressed) ...



pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: [GENERAL] Testers needed ...
Next
From: Tom Lane
Date:
Subject: Re: Index Scans become Seq Scans after VACUUM ANALYSE