[TriLUG] logrotate misery

[Brian A. Henning]

Hi List,
  Every few days, logrotate sends my system into a tizzy.  Sometimes it gets
as bad as pushing the load average above 10, which has the side effect of
causing sendmail to stop accepting connections.  I don't have huge amounts
of logs (this past time, the rotated logs were only up to .3, and I've seen
machines with much heavier log loads have no problems), so I'm not sure what
could be causing logrotate to have these fits.  By the time I really begin
to notice the system's lag, often logrotate has racked up several hours of
CPU time.  I usually end up using killall to send SIGKILL to all logrotate
processes.
  I know there isn't much info here to help reach a diagnosis, but any
guesses?  Samba also seems to have a bit of trouble; when I switch to tty1,
especially if it's been a long time since I've actually used the console
(rather than ssh), there will be a screenful of messages like:

smb_request: result = -104, setting invalid
smb_retry: successful, new pid=17175, generation=285

Could that be related somehow?  It's always the same message, result = -104,
setting invalid.

It's not really a huge deal I guess..  I could always take logrotate out of
cron.daily if nothing else, but I don't necessarily wish to give up the
function that logrotate performs.

Thanks for all ideas tossed my way.

Cheers,
~Brian