FreeBSD Machines dieing, we've tried so much....

[Matt Juszczak]

Hi all,

OK, we're still having the FreeBSD machines die on us.  Its two specific 
machines we've noticed, both FreeBSD 5.4, different hardware, different 
purposes.

Originally, orion, our mail server, started getting kernel traps and 
dieing.  Then, our primary ldap server, a week later, started doing it.  
Now they both are dieing atleast once every couple days, at random 
times.  Orion has been up solid for five days, and Caliban (our primary 
ldap server) has been up for about seven, before this evening at 2:00 am 
when it died again.

Here is the output from Caliban: http://paste.atopia.net/126.  Orion has 
a similar message on the console when it hard locks, but the process 
usually says "procmail".

I've never had instability problems with FreeBSD.  These machines are 
both in the same location, but on different power supplies.  They are 
controlled with high-level Air Conditioning. We've got three other 
FreeBSD 5.4 machines which haven't shown any sign of instability, but 
they dont receive anywhere near as much traffic as Caliban and Orion ... 
those servers get hammered constantly.

The ONLY similarity between Orion and Caliban software-wise is that they 
both are involved in LDAP.  Caliban acts as a primary LDAP server and 
Orion has LDAP configured via pam and nss.

Please let me know any suggestions you can think of. The hardware is 
fairly new in both machines, but they are completely different kinds of 
boxes. Both machines are multiprocessor.

Thanks in advance,

Matt