Timeouts since switching to Swiftel

[gstark]

I'm seeing some evidence of some network timeout issues on my sendmail server. These have only started to happen within the last week - since I switched from IINet to Swiftel.

Here's some examples of what I'm seeing in my logs ...


Aug 19 15:32:21 giulia sendmail[9747]: h7J5VP0F009747: server1.zend.com [209.61.191.11] did not issue MAIL/EXPN/VRFY/ETRN during connection to MTA

Aug 21 08:47:29 giulia sendmail[14422]: h7KMkP0F014422: mail.rhn.redhat.com [66.187.232.134] did not issue MAIL/EXPN/VRFY/ETRN during connection to MTA

Aug 21 10:34:57 giulia sendmail[14557]: h7L0Y90F014557: pix-535-pool.redhat.com [66.187.232.200] did not issue MAIL/EXPN/VRFY/ETRN during connection to MTA

So some incoming connections are now timing out when they previously were not, and they should not now be doing so. The obvious change is in my upstream network and paths, and the settings therein, as well as my new upstream DNS hosts.

I had set the MTU setting on the NIC to 576 because I'd previously had problems with larger files coming in, and that cured that problem, but it's not doing anything for this one. Setting it to 1500 did nothing either except bring back the previous problem, so clearly that's not what I want to do.

Does anyone here know what other issues affect the timing of sendmail and/or DNS responses?

Linux is RH9.0; sendmail is 8.12.8

Thoughts and suggestions, good people ?

[DSL_Tech]

This thread offers the best explanation:

http://forum.swiftdsl.com.au/showthread.php?s=&threadid=1136

[gstark]

Thanx, but that wasn't the issue I was seeing.

I was actually seeing these issues two days before anyone had even heard of Sobig F, and I could, using some external rexources, trigger this problem at will.

But I did find an answer, and for those who may be interested, it was a timeout issue with a couple of the dnsbl servers. They weren't responding within a reasonable time, and were thus causing the resultant issues that I was seeing.

It took a bit of trial and error to fix - chopping and changing the dbsbl servers I was referencing until I found the offender, but it's all fixed and done now.