Damian Menscher said: > On Sat, 14 May 2005, Mark wrote: >>> Damian Menscher wrote: >>> >>> Obviously I always choose the F= option so my mailservers >>> will continue to work in the event of a milter failure. >> >> That is your prerogative. I, on the other hand, would never configure >> sendmail in that fashion. Clamav-scans are a critical part of mail >> delivery. I would never allow mail to be delivered when the virus >> scanner >> is down, and would indeed TempFail delivery until I restarted the daemon >> (via a watchdog script). > > My situation is somewhat unique in that I have intelligent users and > very few windows machines. So viruses are more of an annoyance than a > threat. But let's ignore that for now.... > > Both of us run watchdog scripts (I run clmilter_watch every 15 minutes) > so, at worst, clamav-milter will be down for 15 minutes. In your case, > all mail delivery will stop, and I think outgoing mail will also be > broken. So, not only will your mailserver be down, but your users won't > be able to contact you to let you know it's down. Of course, a > 15-minute downtime might be acceptable to you.
Use a startup script like this (crude) example and you wait only a few seconds: #! /bin/sh # Start daemon, restart if it dies, send notification to syslog start_clamav_milter () { /usr/bin/clamav_milter [args] } # main While :; do start_clamav_milter sleep 5 logger -t clamav -p local5.crit "Damn milter quit again" done You can also run cron with * * * * * watchdog.sh ... and wait a minute at most. In either method you need to keep an eye on the logs and procs remotely but that's what Big Brother or Big Sister is for. dp _______________________________________________ http://lurker.clamav.net/list/clamav-users.html