Mattias Norlander wrote:
>
> Hi!
>
> I've been running postgres for an online sports service for some
> months now. Everything has worked quite smoothly, but this morning
> something strange happend.
>
> Two postgres processess started by the webserver each reached over
> 300M in size and slowed the machine down to a crawl. They had been
> running for a couple of minutes, after i killed them everything
> went back to normal.
>
I once had the same problem, and I tracked it down to a corrupt
database.
Always after starting postmaster, the process quickly went over 300 Mb
and let the machine dye.
After tracking down, which daatabase was the problem, I just put in the
backup
of the last night and the problem disappeared.
I think it's related to corrupt database or index files, but was never
able to
confirm this.
Since then, the postgres user has a reasonable ulimit for the
memory-sizes of
processes...
Cheers
Florian
