3.5 smgr keeps dying

[Marc Weinstein]

My smgr dies every so often and has to be restarted.  I first noticed
the problem when I installed some database software, and this software
started chewing up enormous numbers of in-memory file pointers.  Apparently,
it tried to open more files than the system limit, and right then my
smgr died.  When I restarted the smgr, and re-opened the databases,
it died again.

Since then, it dies even when I haven't touched the database.  I'm not
sure if there's something I broke.  I can't correlate it with anything so
far.

Has anyone seen this problem or knows how to fix it (or prevent it)?

I plan to upgrade to 3.51 on a 3b1, and I'll give myself more file pointers.
I would hope that'll solve the problem...

-- 
Marc Weinstein
{simon,royko,tellab5}!linac!fithp!mhw		Elmhurst, IL
-or- {internet host}!linac.fnal.gov!fithp!mhw