Tek-Tips is the largest IT community on the Internet today!

Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!

  • Congratulations Mike Lewis on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!

Problem with Win2K and TCP/IP connect during long xmit sessions

Status
Not open for further replies.

dovmar

MIS
Jun 10, 2002
5
US
Hello all,

This may not be the right forum, but you gotta start someplace....

I am a small ISP. I have a client who has a large (300K) email list and uses some mailer software to send bulk mail via an install of Imail 8.whatever_is_current . Here's what happens.

The server will move along fine for a few days and then, in the middle of the transmission session, will begin to get 'MX connect fail' errors. This means that the Imail smtp service has gone out to DNS, gotten the address of the recipient mail server, attempted to connect to it, and failed to initiate a session.

As time goes on, over thenext few hours, more and more of these failures occur.

Further, if you then go to a dos-box and try to telnet to the ip address port 25 (as a test), it fails to connect!

Now, these are mail servers for MCI, AOL, Hotmail, earthlink and so on, not just little guys like me with 50 or so domains.

What I am thinking is there is some memory leak, or protocol stack problem that after extended xmission, the nic or the driver just looses its mind. Or some internal DNS cache gets corrupted. It seems odd that servers who take mail and connect at say, 7:00 AM, start 'failing to connect' at say 3:30, but, will, after a reboot, start taking mail and responding appropriately.

Keep in mind also that not all of my transactions fail. Many work. But once this server starts to fail, many more failures of the same type occur, and then I generally reboot the server, and it clears up. Till today.

Other info: server runs IIS 5, sqlserver 2000, coldfusion 6.1 (MX) and is a dell poweredge 1400SC dual cpu unit. I am not using the onboard NIC, I am using a 3com 3c905 series nic.

Any thoughts, or directions?
Marty
 
I've worked a little with I-Mail and have seen similar problems. I just logged onto the I-Mail server, stopped all the services running, then re-started the services and all was well with the world. Had around 250 clients also. Worth a shot to start with. We also have an I-Mail forum here.
forum887


Glen A. Johnson
If you're from Northern Illinois/Southern Wisconsin feel free to join the Tek-Tips in Chicago, Illinois Forum.

TTinChicago
Johnson Computers
 
Glen,
Thanks for a thoughtful post. I have tried what you suggest. What seems to happen is as follows.

The Imail send begins to get MX-connect-fail situations
We stop and restart the queue manager and SMTP service (both Ipswitch programs)
We see the same pattern.
We reboot the system
FOr the first 10 minutes or so we get phenomenal thruput - like 45 to 60 thousand delivery attempts with delivery an hour.
Then, mx connect fail messages
then slow.....to.....c.r..a..w..lllll !!

I keep thinking this is somehow memory related, either in nic memory, tcpip stack memory, or internal dns cache memory... It's just too coincidental that reboots allow really great thruput, then craps out.

Any other thoughts? Anyone?

Thanks again
Marty
 
Have you tried setting up a monitor on the server and see what the tcp throughput is? Actually, a monitor on any of the things you mentioned. Good luck.

Glen A. Johnson
If you're from Northern Illinois/Southern Wisconsin feel free to join the Tek-Tips in Chicago, Illinois Forum.

TTinChicago
Johnson Computers
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Back
Top