Performance failure / automatic restart / cron ?

I have this notorious problem: at some point pgsql simply hangs up and
requires a restart.

---------------------------------------------------------------------------
-------
homer:/usr/share/doc/postgresql-contrib# /etc/init.d/postgresql restart
Restarting PostgreSQL database: postmaster
Stopped /usr/lib/postgresql/bin/postmaster (pid 14243 14245 14247 14750
14811 1489 15569 15570 15573 15584 15643 15644 15665 15800 15801
15804 15817 15874 16072 20473 20474 20549 20763).
Starting PostgreSQL postmaster.
postmaster successfully started
...
homer:/usr/share/doc/postgresql-contrib#
---------------------------------------------------------------------------
-------

Whenever i restart the server i get a long list of pid's - sometimes its many
lines.

I wrote a little script to restart, vacuum and analyze the database. And
although the script runs, there are no results. i have to restart the db
manualy...
ive seen a post about vacuum an analyze not working from the cron,
unfortunately there was no solution.

Ill be happy with any ideas - in the last week the db required me to have it
restarted once a day which is not very convenient.

I realize my app is very resource hungry but the thing is that it works great
with mysql. The problem occurs only with pgsql which i realy need.

Best regards,
Piotr