| |||||||||||||||||||||||||||||||
|
> From: owner-postfix-users > [mailto:owner-postfix-users] On Behalf Of Victor Duchovni > Sent: Friday, 9 May 2008 4:02 AM > To: postfix-users > Subject: Re: mailq lockups > > On Thu, May 08, 2008 at 01:41:06PM -0400, Joey wrote: > > > > No it should be "pipelining", because this is a *discard* keywords list. > > > > To keep it simple I took out the pipelining statement as a quick test, > > and that didn't work. > > How did you do that? What evidence did you collect to check > that PIPELINING was not used? > > > Is there something from the spamassassin/ anti-spam side that could be > > a factor in this? > > How could this possibly matter? Postfix delivers mail long > after all the scanning is done. Delivery is just pumping bits > out to the network. > > > The exchange server has NOTHING for anti-spam, the postfix server acts > > as a gateway to accomplish this. > > Don't make random guesses, measure! How's that "tcpdump" going??? > > -- > Viktor. You can also measure on the Exchange side by examining the SMTP logs (which you should have, despite the fact they're not enabled by default) and also the Application event log on the Exchange server. You can also increase the verbosity of the application logs on the Exchange server on a per-service basis. This KB article gives some elementary steps: http://support.microsoft.com/kb/257265 - you want to at least monitor the Transport service (and log SMTP, as I mentioned). Also, do you have Outlook installed on the Exchange server (which mucks up MAPI) or the SMTP service enabled in IIS on an Ex2K7 transport server? Both of those are no-nos.
| ||||||||||||||||||||||||||||||
© 2004-2008 readlist.com