Auteur: Daniel Buchholz Date: À: Exim-Liste Sujet: [Exim] PANIC! Do I need more memory?
Hi!
Some days ago we had serious trouble with our exim machine apparently having
trouble with memory allocation. (Of COURSE while I was not in the office)
Maybe sombody on the list can help me in finding out what the problem was.
First of all some log entries:
warn:
Apr 19 12:43:28 mailer inetd[126]: fork: Cannot allocate memory
Apr 19 12:43:52 mailer last message repeated 7 times
Apr 19 12:44:30 mailer last message repeated 11 times
Apr 19 12:44:35 mailer inetd[126]: accept (for pop3): No buffer space
available
Apr 19 12:44:35 mailer last message repeated 331 times
Apr 19 12:44:35 mailer inetd[126]: fork: Cannot allocate memory
Apr 19 12:44:36 mailer inetd[126]: accept (for pop3): No buffer space
available
...
messages:
Apr 19 12:43:28 mailer inetd[126]: fork: Cannot allocate memory
Apr 19 12:43:52 mailer last message repeated 7 times
Apr 19 12:44:30 mailer last message repeated 11 times
Apr 19 12:44:35 mailer inetd[126]: accept (for pop3): No buffer space
available
Apr 19 12:44:35 mailer last message repeated 331 times
Apr 19 12:44:35 mailer inetd[126]: fork: Cannot allocate memory
Apr 19 12:44:36 mailer inetd[126]: accept (for pop3): No buffer space
available
Apr 19 12:45:00 mailer last message repeated 1708 times
Apr 19 12:45:00 mailer /usr/sbin/cron[168]: (CRON) error (can't fork)
Apr 19 12:45:01 mailer inetd[126]: accept (for pop3): No buffer space
available
...
exim_paniclog:
2000-04-19 12:44:41 50 accept() failures: No buffer space available
(repeated up to 10 times per second for up to 3-4 minutes,
same in mainlog)
The only thing I can say is that just before the problem occured exim
received 5 messages with about 25 recipients each and message sizes from 32k
to 104k.
remote_max_parallel was set to 20 at that time so I think this was the cause
for the problem (I reduced to 5 now). The machine has 64 Megs total. Load is
VERY moderate I think. April average was just below 1000 messages per day.
