I'm currently running exim 4.40, since I've switched to linux kernel
2.6.7 (2.4.x before) exim sometimes misses an alarm, so the queue
runners do not start. When sending the daemon a SIGALRM via kill,
everything works fine for one day, or two. Does anyone else encounter
these problems?
As exim reacts to a manual SIGALRM, either the original SIGALRM gets
lost somewhere (kernel failure?) or exim does not set alarm at all. The
only possible reason for the later, as I could see from browsing the
sources, would be when someone uses an expansion in
smtp_accept_max_per_host, that in effect uses its own alarm (like ldap
lookups, or readsocket). But I don't use expansions in
smtp_accept_max_per_host.
Nico
P.S. the same happened with exim 4.34, so it's not 4.40 specific.
