Am am getting a lot of bayes locking errors as well. Probably related. Should the configuration options listed below help with bayes locking issues in a high-volume situation?
--Aaron
Lucas Albers wrote:
It might be baye-learning is making mimedenfag make too many slaves. I encountered this situation, but in my case it just stopped spawning slaves.
#http://mail.stalker.com/Lists/CGatePro/Message/52889.html bayes_learn_to_journal 1 bayes_journal_max_size 5120000
SA will get a lock then Mimedefang will spawn a new slave, which can't get a lock, which spawns a new slave,etc.
Aaron Paetznick said:
MIMEDefang has been crashing on me fairly regularly and it's driving me bonkers. Basically it goes "to error state" and doesn't recover. I have a script watching the logfile, and when there is a problem MD gets automatically restarted, but when this happens it won't start up again until I manually remove the /var/lock/subsys/mimedefang file. Here's a clip from /var/log/maillog from when this last happened. I would consider increasing the children, except my MX_MAXIMUM is already set to 80. I just now increased the MX_REQUESTS to 500, but I'm not confident that that will help too much. Any ideas? This is MD v2.41, SA v2.63 + the usual suspects on RHEL 3.0 with perl v5.8.3. Thanks.
_______________________________________________ Visit http://www.mimedefang.org and http://www.canit.ca MIMEDefang mailing list [EMAIL PROTECTED] http://lists.roaringpenguin.com/mailman/listinfo/mimedefang
