From: | Jeff Boes <jboes(at)nexcerpt(dot)com> |
---|---|
To: | pgsql-admin(at)postgresql(dot)org |
Subject: | Linux 2.4.x VM |
Date: | 2001-11-21 22:40:19 |
Message-ID: | 20011121174019.4399b140.jboes@nexcerpt.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-admin |
We recently experienced a situation that may be of value to others.
Our installation was Postgres 7.1.2, Linux 2.4.9. We had a bug in a
process that was essentially causing it to do a 'select * from table'
where the table was large, every once in a while. This would cause
the process's memory requirements to grow very large, and eventually
this would overwhelm the database by corrupt the shared memory (we
think this happened when the OS's virtual memory system fell behind
sufficiently). Sometimes, we'd even see corruption in tables.
Now we're at Linux 2.4.14 and Postgres 7.1.3. Before we found the bug
in the process, we would see one or two of the processes grow to a
gigabyte or so of memory, and then they would get killed by the OS.
However, we wouldn't see the shared memory corruption.
Our conclusion is that Postgres 7.1.2/3 and Linux 2.4.(<14) is not a
stable configuration. If you are running an old kernel you may wish
to upgrade soon.
--
Jeff Boes vox 616.226.9550
Database Engineer fax 616.349.9076
Nexcerpt, Inc. jboes(at)nexcerpt(dot)com
From | Date | Subject | |
---|---|---|---|
Next Message | Martins Zarins | 2001-11-22 09:37:47 | indexes |
Previous Message | Björn Lundin | 2001-11-21 16:31:32 | Re: Problems between Mandrake and pg |