On Mon, Jan 5, 2009 at 6:17 AM, Ygor Degani <ygordegani@gmail.com> wrote:
> I use postgres-8.3.5. My database has 168 GB in size. Whenever i try to do
> vacuum full happens the following error:
>
> WARNING: terminating connection because of crash of another server process
> DETAIL: The postmaster has commanded this server process to roll back the
> current transaction and exit, because another server process exited
> abnormally and possibly corrupted shared memory.
> HINT: In a moment you should be able to reconnect to the database and
> repeat your command.
> vacuumdb: vacuuming of database "acotel_brazil_vas" failed: server closed
> the connection unexpectedly
> This probably means the server terminated abnormally
> before or while processing the request.
>
> If i have autovacuum enabled, Do i need to do vacuum full?
You have two problems. One is you're running vacuum full regularly,
which you shouldn't need, and which can cause bloated indexes. Normal
autovacuum should handle your database fine unless you're deleting a
huge part of a table over and over.
As to why it's crashing, that's another story. Anything in the system
logs about the oom killer kicking in? Have you got work_mem set to
something huge? maintenance work mem really huge? shared_buffers,
etc? What have you changed from stock in postgresql.conf?
