CentOS 5 / SendMail 8.14.x
I'm noticing intermittent errors with a milter on my SendMail server. Here's an excerpt from the maillog (I've edited hostname, QID, milter name, and recipient address to protect the innocent :-) ):
Jul 3 4:11:10 FOOBOXHOST foomilter: ABcDEFgHI12345: MAIL FROM:<>
Jul 3 4:11:10 FOOBOXHOST foomilter: ABcDEFgHI12345: msg_size='12345'
Jul 3 4:11:10 foomilter: ABcDEFgHI12345: RCPT TO:<mailbox@fooexample.com>
Jul 3 4:11:11 FOOBOXHOST foomilter: foomilter: mi_stop=1
Jul 3 4:11:11 FOOBOXHOST foomilter: exiting foomilter
Jul 3 4:11:11 FOOBOXHOST sendmail[11111]: ABcDEFgHI12345: Milter (foomilter): write(D) returned -1, expected 6: Broken pipe
Jul 3 4:11:11 FOOBOXHOST sendmail[11111]: ABcDEFgHI12345: Milter (foomilter): to error state
Jul 3 4:11:11 FOOBOXHOST sendmail[11111]: ABcDEFgHI12345: Milter: helo=upstream.fooexample.com, reject=451 4.3.2 Please try again later
Jul 3 4:11:11 FOOBOXHOST sendmail[11111]: ABcDEFgHI12345: Milter (foomilter): write(D) returned -1, expected 86: Broken pipe
Jul 3 4:11:11 FOOBOXHOST sendmail[11111]: ABcDEFgHI12345: Milter (foomilter): to error state
What's does write(D) returned -1, expected 6: Broken pipe mean?
I find this line particularly interesting:
Jul 3 4:11:11 FOOBOXHOST foomilter: foomilter: mi_stop=1
Asking Mr. Google about it took me here: http://www.mailshell.com/mail/client/oem2.html/step/sendmailfaq#25 which had a blurb about the server being under heavy load but when I check basic maillog info, I'm not seeing particularly high numbers (a couple thousand messages an hour... pretty evenly distributed).
Any ideas about what could be going on here and how to prevent it from occurring?
