Postfix and Queues

[Vladimir Dvorak]

Vladimir Dvorak wrote:

> Hi *,
>
> I have little, but not annoying problem with Postfix and queuing 
> management. When two or more big (understand >2 MB) emails are in 
> active queue, Postfix is trying to deliver them simultaneously. But 
> moth ago began to apear following message in /var/log/mail :
>
> 5BCED8EFE3* 4534286 Fri Mar 25 14:25:35  user at mydomain.cz
> (lost connection with mailserver.domain.com [aa.bb.cc.dd] while 
> sending message body)
>
> ^^ this message is in defer queue and wait for its time, when next try 
> to deliver comes. During this time comes another big email from local 
> user, which get stucked in defer queue too. The link start to have 
> very high load of outgoing mail traffic, but all of them are 
> unsuccesfully delivered.
>
> Well, is there any chance to reduce queue to one mail in active queue ?
>
> The system runs postfix-19991231pl08-29 - I know its very old.
>
> Thank you in advance.
>
>

I found solution.

There is (in front of this mailserver) OpenBSD firewall with "scrub in 
all". Problem appears, when this messages are targeted to Microsoft MX. 
MS servers badly handle with IP headers and sets "Dont fragment" bit, 
when this packet is fragmented. :-)

Excerpt from man pf.conf:
"Some operating systems are known to generate fragmented packets with 
the dont-fragment bit set. "

If you want to avoid this problem, leave "scrub in all" and use "scrub 
in no-df".

If you have another solution, you are welcome! :-)

 Vladimir Dvorak
-- 
* VDSOFT.ORG           dvorakv at vdsoft.org      *
* (+420) 602 944 941     http://www.vdsoft.org *