Intermittent telnet smtp 25 failures... 1

[xedgex]

We have an application on a Windows 2003 server which has been able to send emails back to a view once an action takes place on the website. This broke entirely altogether, I've begun to troubleshoot the network layer aspect and by running telnet destination-ip-address 25 I get a valid response about 50% of the time, the other 50% of the time telnet over port 25 fails with the generic error of;

Could not open connection to the host, on port 25: Connect failed

I can ping this IP address always, even when these telnet tests fail. When they work I can send test emails through them. I can tracert when it works and doesnt work.

I have tried other destination IP addresses as the SMTP test, to known good sources, these work in conjunction with the original IP address that I test or they both fail.

Our firewall doesn't prevent port 25 traffic from inside > outside. The AV software was disabled during my testing. No windows firewall.

The last significant change was that this server was clustered with another, then clustering was broken off. Another change was that the NIC cards were teamed, as this is an HP server.

Any helpful suggestions are greatly appreciated. Thanks for your time!

 

[VinceWhirlwind]

Disconnect the server you are trying to telnet to from the network and then try pinging the IP address again.

 

[pompeychimes]

First rule of troubleshooting teamed nics is to diable the teaming and test again. Once de-teamed clear arp entries and test again.

Vince's speculation (I assume) that you may have a dup IP could also be the result of a misconfigured team.

James

 

[burtsbees]

Telnet tests up to the application (layer 7), all the way through all 7 layers...

/

tim@tim-laptop ~ $ sudo apt-get install windows
Reading package lists... Done
Building dependency tree
Reading state information... Done
E: Couldn't find package windows...Thank Goodness!

 

[xedgex]

Unfortunately the server that I am trying to telnet to over port 25 is a shared machine not in the same facility, I can not disconnect it.

I also just found out this server was part of a cluster, which was then broken up into stand alone servers, the issue may have surfaced during the de-clustering task but I am not 100% sure.

If you feel that disabling the teamed NICs may help I will visit the server and accomplish this, please advise.

Thanks for all the feedback!

 

[pompeychimes]

Are there any static routes configured on the Server? IF unsure post the "route print" output here. If i understand correctly you're initiating the pings and telnets from a device on a different IP network than the server. Do you get the same results when pinging and telneting from a device on the same IP network as the Server? Also, when the testing fails issue the "netstat" command on the Server to see if a connection attempt even reaches the Server.

 

[xedgex]

I am initiating the telnet port 25 test from my own server to a remote server. When these telent tests fail off my server, they do NOT fail from differrent locations such as my laptop or other workstations. Also when these telnet tests fail, PING tests DO NOT fail.

I checked for static routes, and there are none set.

Furthermore, I can always telnet to the destination server over port 25 from other servers on the same LAN, just not this one particular server when it fails.

I've checked firewall logs, and nothing seems to be blocking it at that level.

 

[xedgex]

Using netstat, when the telnet port 25 connection fails, this is the netstat line;

TCP DODGER:3848 gate.sat.emailsrvr.com:smtp SYN_SENT

When it works netstat line is this;

TCP DODGER:4151 gate.sat.emailsrvr.com:smtp ESTABLISHED

 

[burtsbees]

Throw Wireshark on the server, run it, then try. I'll bet you never get a SYN-ACK and you get a ton of stripes (dropped packets due to excessive round-trip time, latency and/or dup acks). The server you are trying to telnet to is problematic, I would assume.

/

tim@tim-laptop ~ $ sudo apt-get install windows
Reading package lists... Done
Building dependency tree
Reading state information... Done
E: Couldn't find package windows...Thank Goodness!

 

[xedgex]

Solved. It turns out the issue was caused by the HP Team NIC setup. When I unpaired them, one of the interfaces was setup for a wrong internal IP subnet. Running the server with a single NIC, no teaming, working 100%. Thanks for all the suggestions.