ibarrac@serss0.fiu.edu (Carlos Ibarra) (12/01/90)
We have been experiencing a problem with our mail system which I would like to solve. We have noticed a couple of times that when our main mailhost machine is rebooted, some people get mail messages sent from that machine which have been in the mail queue for a week or two. This is not acceptable; electronic mail should be faster than that. I believe I found the cause of the problem. When sendmail processes a mail job, it creates a lock file in /var/spool/mqueue which prevents other invocations of sendmail from attempting to process the same mail job. If sendmail dies while processing the locked job, the lock file remains and prevents the following queue runs from processing the job (sendmail currently processes the queue every hour), until the machine reboots. A reboot causes rc.local to clear the lock files before restarting sendmail. I am not sure why our sendmail died or somehow failed to delete the lock files. I will attempt to find out. A possible solution might be to have cron run a script every night which will kill all sendmail processes, delete the lock files, and restart sendmail. Can anyone suggest a better solution? We are running SunOs 4.0.3. Carlos A. Ibarra ibarrac@fiu.edu