Measuring activity/problem with timeouts

[Rob Tanner <rtanner@linfield.edu>]

I'm in the process of phasing in an OpenLDAP v2.2.17 (replacing an old
Netscape 4 server), and every night, around midnight and then again at around
4:30am, the mail system logs a bunch of timeouts accessing the LDAP server --
timeouts both when binding and searching.  The rest of the time, everything
appears to be running fine.

I've up'd the logging level to 286 looking for issues on the OpenLDAP server,
but nothing appears to be wrong.  The only measure of activity I have
currently is to count log entries, and between 4:39:00am and 4:39:59am, a
time when the mail system registered timeouts accessing OpenLDAP, there were
221 log entries.  A similar count at at 9:10am showed 168 log entries.  The
mta (Postfix) shows a similar pattern in log entries (counts 72 and 64
respectively).  But in my mind, these numbers don't really indicate a lot of
activity.  Nor is the problem strictly an issue with the mta since nagios, a
system services monitoring tool, likewise reports timeouts.

What I need to so is figure out what's causing the timeouts.  Any suggestions
as to how, or at least how to measure what the ldap server is really doing.


Thanks.

-- 
Rob Tanner
UNIX Services Manager
Linfield College, McMinnville OR