From: Igor Brezac (no email)
Date: Tue Jun 18 2002 - 10:17:35 EDT
Solaris unix sockets are not very solid. If you have high mail traffic,
it is possible that you are hitting this os bug. Try to listen on
On Tue, 18 Jun 2002, Kiarna I Boyd wrote:
> Thanks Michael!
> I did stop(&pkill) and restart the master before rebooting, didn't seem
> to help.
> I will look into the prefork number.
> Any recommendations on that?
> lmtpunix cmd="lmtpd" listen="/var/imap/socket/lmtp" prefork=1
> Quoting Michael Bacon <>:
> > --On Monday, June 17, 2002 07:10:21 PM -0500 Amos Gouaux
> > <+> wrote:
> > >>>>>> On Mon, 17 Jun 2002 17:27:29 -0400,
> > >>>>>> Kiarna I Boyd <> (kib) writes:
> > > kib> Jun 15 13:05:49 thoth sendmail: [ID 801593 mail.crit]
> > > kib> g5FH5mnl009941: SYSE
> > > kib> RR(root): Could not connect to socket /var/imap/socket/lmtp:
> > > Connection kib> refused
> > > kib> This error repeats until it is the only thing in the log.
> > >
> > > That seems odd. Then again it has been a while since I've messed
> > > with Sendmail. Sorry.
> > You obviously never got bit by the rabid database lockup monkeys.
> > The
> > problem here isn't sendmail, it's that lmtpd stopped responding, for
> > reasons unknown. Before we moved off of the Berkeley DB format under
> > 2.0.16, and even more back on 2.0.11 and 2.0.14, this was a regular
> > occurance.
> > Rebooting probably isn't necessary. Stopping and restarting the
> > master
> > daemon will do the trick most of the time, as disruptive as it is.
> > You can
> > also try increasing the prefork number on lmtp in cyrus.conf, and HUP
> > the
> > master. If that works, you're looking at some variation of the
> > process
> > accounting problem, which I don't think has been fully solved yet.
> > Michael