I've seen this before during periods of network congestion. You could
increase the max connections and see if that helps. In my case it helped
for a short period of time. What I ultimately ended up doing was deploying
two dedicated IMS servers and split the smtp traffic between mailbox
servers, using address spaces limited to location.
Are there any applications on your internal network using your exchange box
to relay mail, that might possibly be tying it up? Possible insider
spamming or legitimate mass mailing? This is just a long shot.
> We're using a Watchguard firewall. The equivalent of a static conduit is
> place. SMTP sessions can be established with remote sites using different
> mail software from a different machine.
> This is just a guess, but it almost appears that the allowed number of
> outbound connections on the server is being exceeded without the
> sockets ever being released. That would explain why rebooting the mail
> server and only the mail server corrects the problem. If thats the case,
> doesn't it happen on a more regular basis?
> > Is the firewall a PIX, and is mailgaurd enabled if it is? You do have a
> > static conduit setup for the IMS, right?
> > --
> > Michael Abbaticchio
> > http://www.abbaticc.com
> > http://www.abbaticc.com/exchange
> > > Yes, it is behind a firewall running NAT. What has me confused is that
> > > other protocols work properly and even SMTP runs more often than not.
> > > viciousdogs
> > > > If you mail server inside of a firewall? If so, does it use a
> > > > outside address, or does it go out via NAT pool?
> > > > --
> > > > Michael Abbaticchio
> > > > http://www.abbaticc.com
> > > > http://www.abbaticc.com/exchange
> > > > > I am running Exchange 5.5 SP4 on under WinNT 4.0 SP6a. The
> > > server
> > > > > is only set up to run Internet Mail Services. The problem I am
> > is
> > > > > that the Internet Mail Connector randomly loses its ability to
> > > > > connections to outside servers. Connections to other internal mail
> > > servers
> > > > > works fine.
> > > > > The event logs report an error 3010 with a description of "An
> > to
> > > > > connect to host <hostname> failed." The connection to the DNS
> > > fine.
> > > > > Once this happens no messages can be sent until the exchange
> > > > > rebooted. Shutting down the Exchange services and restarting them
> > no
> > > > > effect.
> > > > > The problem pops up anywhere from every 45 minutes to 10 days. No
> > other
> > > > apps
> > > > > are running on the mail server. Incoming mail still works fine.
> > logs
> > > > > leading up to the first event never show any patterns as to what
> > > be
> > > > > causing the problem.
> > > > > Anybody have any ideas?