Command time limit exceeded: "/usr/lib/cyrus/bin/deliver"

[Laurens van Alphen]

Hi all,

I'm having some trouble with a Cyrus box. The box is running Postfix as MTA.
Cyrus uses saslauthd to authenticate using pam, which in turn uses
pam_mysql.so to authenticate against a MySQL database.

This is now the second time that, after a few days uptime, I get things like
this:

Aug 22 20:32:49 hostname postfix/qmgr[pid]: 9728391B: from=<from at address>,
size=5648, nrcpt=1 (queue active)

And then 1001 seconds later:

Aug 22 20:49:29 hostname postfix/pipe[pid]: 1234567A: to=<one at two.three>,
orig_to=<two at three.four>, relay=cyrus, delay=1001, status=bounced (Command
time limit exceeded: "/usr/lib/cyrus/bin/deliver")

The messages is then bounced (not deferred) by Postfix, a real pain.
Postmaster gets the following explaination:

The e-mail system was unable to deliver the message, but did not report a
specific reason. Check the address and try again.  If it still fails,
contact your system administrator. < hostname.domainname #5.0.0 X-Postfix;
Command time limit exceeded: "/usr/lib/cyrus/bin/deliver">

>From the Postfix source code:

        if (pipe_command_timeout) {
            vstring_sprintf(why, "Command time limit exceeded: \"%s\"%s%s",
                            args.command,
                            log_len ? ". Command output: " : "", log_buf);
            return (PIPE_STAT_BOUNCE);
        }

The Cyrus logfile shows nothing but:

Aug 22 20:32:49 hostname master[pid]: about to exec /usr/lib/cyrus/bin/lmtpd
Aug 22 20:32:49 hostname lmtpunix[pid]: executed

What I do know about this problem is that (just like the previous time) a
restart of saslauthd and cyrus fixes everything (for the time being). Just
restarting Cyrus does not fix the issue.

Hints are more than welcome!

Thanks in advance,

Laurens van Alphen