Connection Reset by host --smtp error-local only

[billnoble]

We have 2 mail servers,one sendmail(8.9.3 on linux(debian)and an Exchange Server5.5 sp4.
These servers were previously on two separate networks (each in their own DMZ) off a pix 520. We forward some mail from the sendmail server to accounts on the exchange server. While these servers were separated by the pix, they worked together without problem.
We recently re-ip'd our network and the two servers are now on one network in the same dmz.
Now when these machines try to talk smtp to each other the connection is reset by the contacted host.

TCPDUMP from the linux box is as follows:

tcpdump: listening on eth0
15:16:16.767042 newmail.ocl.net.3530 > mail.ocl.oxford.on.ca.smtp: P 2435498689:2435498695(6) ack 1859690812 win 32120 (DF)
15:16:16.767200 arp who-has newmail.ocl.net tell mail.ocl.oxford.on.ca
15:16:16.767241 arp reply newmail.ocl.net is-at 0:1:3:dd:42:f8
15:16:16.767341 mail.ocl.oxford.on.ca.smtp > newmail.ocl.net.3530: R 1859690812:1859690812(0) win 0
15:18:09.500590 newmail.ocl.net.3609 > mail.ocl.oxford.on.ca.smtp: S 33661534:33661534(0) win 32120 <mss 1460,sackOK,timestamp 9661739[|tcp]> (DF)
15:18:12.495249 newmail.ocl.net.3609 > mail.ocl.oxford.on.ca.smtp: S 33661534:33661534(0) win 32120 <mss 1460,sackOK,timestamp 9662039[|tcp]> (DF)
15:18:14.495240 arp who-has mail.ocl.oxford.on.ca tell newmail.ocl.net
15:18:14.495347 arp reply mail.ocl.oxford.on.ca is-at 0:50:4:b0:f2:8b
15:18:18.495238 newmail.ocl.net.3609 > mail.ocl.oxford.on.ca.smtp: S 33661534:33661534(0) win - <mss 1460,sackOK,timestamp 9662639[|tcp]> (DF)

It appears as though in this case the exchange server is reseting the connection.
But on subsequent retrys, they actually connect and tranfer the mail as seen above.
these boxes ping fine and work perfectly with the inside of our network and the rest of the world.
we have replaced the switches, wires, nics, gone from switch to hub, hub to switch, reinstalled tcpip on the exchange server...
any assistance would be appreciated..
bill

 

[GiaBetiu]

I don't know exactly. For me it looks like a problem in IP/Name resolution.
Check the DNS configurations. Be sure that there are not such problems.
One of the difference that is now, comparing with the initial configuration is that they are in the same network. So broadcasts, ARP queries, name resolutions should be consistent (two dissimilar systems!).
Please tell me more about. If I will find more informations I will announce you.
Success! Gia Betiu
m.betiu@chello.nl
Computer Eng. CNE 4, CNE 5

 

[billnoble]

no, both boxes resolve fwd and rev fine from the inside, and from the real world and from each other..
reviewed all dns settings...no problem there.
we are currently in the process of setting up a new exchange box for testing purposes. any other ideas in the meantime appreciated