Thanks for the quick and helpful response, Matthias. Works like a
charm!
I took the sendmail options straight from the advice in the fetchmail
man page. Yeah, I should have dug into sendmail also, but it
would be nice if possible for fetchmail to save users that burden.
Are there any circumstances in which fetchmail should call
sendmail *without* -i? Does fetchmail ever insert a . on a line by
itself to indicate the end of a message to sendmail? If it did,
wouldn't sendmail again give a SIGPIPE error?
Are there other popular sendmail options that could be mentioned?
An update to the fetchmail man page and FAQ would be very helpful.
Cheers,
-Neal
On Tue, Apr 16, 2002 at 01:14:00PM +0200, Matthias Andree wrote:
Neal McBurnett <neal(_at_)bcn(_dot_)boulder(_dot_)co(_dot_)us> writes:
defaults mda "/usr/sbin/sendmail -oem -f %F %T"
Read man sendmail(8) and look for "-i".
It also leaves zombie sendmail processes ala:
neal 18307 0.0 0.0 0 0 ? Z Feb06 0:00
[sendmail <defunct>]
Looks like fetchmail doesn't call wait() to collect it.
It seems like somewhere along the line the "." is not being
escaped properly, as I believe it needs to be for SMTP.
No, all is fine, you just need to configure your sendmail properly.
--
Matthias Andree