[net.mail] Why does sendmail sometimes screw up a lock?

roy@phri.UUCP (Roy Smith) (05/07/86)

	Every once in a while, our sendmail (version 4.12, running on a
4.2bsd Vax-11/750) seems to forget about a message.  The symptom is always
the same; the message hangs around in /usr/spool/mqueue with (I think) qf,
df, xf, and lf files.  Last time this happened, I removed the lf file and
it seems like the message was delivered properly after that.  Our sendmail
logs show:

	May  5 19:02:55 localhost: 29109 sendmail: AA29109: message-id=<8605052302.AA29109@phri.UUCP>
	May  5 19:02:56 localhost: 29109 sendmail: AA29109: from=roy, size=942, class=0
	May  5 19:11:14 localhost: 29136 sendmail: AA29109: locked
	[...]
	May  6 15:11:34 localhost: 4504 sendmail: AA29109: locked
	May  6 17:11:40 localhost: 5422 sendmail: AA29109: to=allegra!ihnp4!sdcrdcf!usc-oberon!blarson, delay=22:08:47, stat=Sent

	So, why do the log files get wedged?  We're a uucp-only site, so
it's not something like a network interface being down.  Actually, I don't
think it has anything to do with uucp at all because sometimes it happens
with local mail.

	We do occasionally run out of swap space.  Is it plausable that
when sendmail tries to fork/exec the delivery agent (uux or whatever) and
this fails because there is no swap space left, the message gets lost?

	If I left things alone, would the lock files eventually time out on
their own?  Is it safe to manually remove the lock after a message has been
in the queue for a few hours?
-- 
Roy Smith, {allegra,philabs}!phri!roy
System Administrator, Public Health Research Institute
455 First Avenue, New York, NY 10016