Connecting process to PostgreSQL works stunningly slow

Hi,

We have a problem with PostgreSQL 8.1 on a production server (Windows
2003 Server, Xeon 3GHz, 1GB)); application placed on it run for a month
without problems. It was great. But yesterday...

Any connection to the server, from application or pgAdmin (both
previously working fine), is slow. I mean, I have to wait *minutes* for
even login to succeed. Even when connecting from the same machine, with
pgAdmin, with no users connected to it. Restarting the service is slow,
opening the database is slow, backup is slow. Sometimes minutes, I
mean. Like there was something wrong with listening for connections -
sometimes I got an information that server on [my_ip_here:my_port_here]
is not listening. The machine runs other services, but there is no
heavy load on it and users do not have any problems with e.g. (don't
shoot) MS SQL Server.

Since I use IPs only, DNS issues should not be involved.

There were no changes to current configuration, autovacuum is on and it
seems to work, according to the log (runs every 60s). There are two
tiny tables (no more than a hundred rows each, barely a database),
there are only updates on them (no more than a few hundreds a day), I
know it's weird but that's how it should work.

I googled and searched on Groups. I'm rather new with PostgreSQL and
dealing with the crashed production server is not nice :)

Tommorrow I will try connection without md5 and look if the drive is OK
(defragmentation?), but maybe someone has any idea, what else could I
check? It is very depressing, since the database is so tiny... If
nothing else helps, I will of course pg_dump, reinstall PostgreSQL
server and wait for a month or so *gulp*.

Thanks in advance,
Bart Golda