RE: Expectations of MEM requirements for a DB with large tables.

From: "Michael Miyabara-McCaskey" <mykarz(at)miyabara(dot)com>
To: "'KuroiNeko'" <evpopkov(at)carrier(dot)kiev(dot)ua>
Cc: <pgsql-general(at)postgresql(dot)org>
Subject: RE: Expectations of MEM requirements for a DB with large tables.
Date: 2000-11-06 06:54:53
Message-ID: 000c01c047be$77da56d0$aa00a8c0@ncc1701e
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general

Kuroi,

Thank you for the response.

I was not specifically thinking of diagnosing my crash, but if you are
curious...

- about a 1000 error messgaes saying that kswapd could not do something
relating to allocating memory.

- after stopping the "psql" job, it still took about 20 minutes before my
Linux OS recovered enough to reboot.

- yep, used "psql", plan on using Perl-DBI soon... I guess from your
response I better make sure that whatever I use never does a full blown
'select *' on that table.

- 2.2.17 from kernel.org, w/RAID patch.

- PostgreSQL 7.0.2 was installed via the RPM from RedHat.

- 98MB of RAM (this is a simple test system only)

- 256MB of SWAP on a Linux SWAP Partition.

- SWAP settings? Hmm... sorry this is where my novice level shows its
colors, didn't know I could modify it.

- Other processes, sure tons of them...

Anyway, based upon your response, I understand that the 1 to 1 ratio is what
crashed my system.

Hmm, now you have me wondering though... I recall reading something that
mentioned that data storage size in PostgreSQL is about 6 times the size of
standard text, so, if my table size is about 5.5GB on the disk does this
mean I really could only need about well say... 5.5GB / 6 = 936MB of RAM?

Any idea?

-Michael Miyabara-McCaskey

> -----Original Message-----
> From: pgsql-general-owner(at)postgresql(dot)org
> [mailto:pgsql-general-owner(at)postgresql(dot)org]On Behalf Of KuroiNeko
> Sent: Sunday, November 05, 2000 9:54 PM
> To: pgsql-general(at)postgresql(dot)org
> Subject: Re: [GENERAL] Expectations of MEM requirements for a DB with
> large tables.
>
>
>
> Mike,
>
> > Anyway, I crashed my system the other day when I did a
> "select *" from
> > one
> > of my large tables (about 5.5gb in size).
>
> Well.... It takes abit more than that to actually crash
> the system. Can
> you give more details? What _exactly_ happened? Did it
> hang? Kernel
> panicked? Something else.
>
> > Does anyone know what the ratio of data output size (say
> from a select)
> > to
> > the amount of
> > RAM used is?
>
> It simply can't be other than 1 : 1. Well, in a sense. Did
> you use psql?
> Other app? If the latter, what interface: API, DBI, other?
> Was it a local
> or remote request?
>
> > Other relevant info, RedHat Linux 2.2.17 kernel, PostgreSQL 7.0.2.
>
> Is this a genuine kernel, or the one shipped by RH? Did
> you build PGSQL
> from sources or was it a pre-built binary? How much RAM do
> you have, how
> fast and big is your swap? Swap settings? Other processes running?
> As you see, there are more questions than answers at this
> stage. Tell us
> more. Tuning the system has never been easy.
>

In response to

Browse pgsql-general by date

  From Date Subject
Next Message Peter Pilsl 2000-11-06 11:00:16 restore blobs (xinv-files)
Previous Message Michael Miyabara-McCaskey 2000-11-06 06:41:12 RE: Expectations of MEM requirements for a DB with large tables.