On Thu, Feb 05, 2004 at 07:12:23PM +0700, Frank Rabitsch wrote:

...

> Difficult to match the failing messages on the sending side with log entries
> on the receiving side. Let me give it a try...
> 
> This is how a normal session should look like according to me:
> 
> 2004-02-03 20:01:07.186652500 tcpserver: pid 23973 from 10.2.0.1
> 2004-02-03 20:01:07.189260500 tcpserver: ok 23973
> mail.ph.magnusint.com:10.2.0.7:628 :10.2.0.1::59579
> 2004-02-03 20:01:07.196407500 ok 1075813267 qp 23974 ddc saved 43 percent
> 2004-02-03 20:01:07.196737500 tcpserver: end 23973 status 0
> 
> It starts up the connection then tcpserver reports OK, and the pid is ended
> in the 4th line. This logentry is from a null client on the firewall on the
> same subnet so on a fast network connection. On the 3rd line it reports ok
> but I don't know how to match the number 23974 to the original message pid
> 23973. If the session is fast its most likely one number higher but on slow
> sessions the number is not sequential.
> 
> The following entries are of a failing session:
> 
> 2004-02-03 18:37:15.950854500 tcpserver: pid 23576 from 10.1.0.7
> 2004-02-03 18:37:15.952341500 tcpserver: ok 23576
> mail.ph.magnusint.com:10.2.0.7:628 :10.1.0.7::32940
> ...
> 2004-02-03 19:37:15.943964500 tcpserver: end 23576 status 28416
> 
> As you can see there is one hour delay and the status code is 28416. Don't
> know what this code means.
> 

status is the exit status. This is what wait(2) returns for the process.
In this case it is an _exit(3) code of 111. This is definitifly the 1h
alarm timer that went off.

> If I do a flush on the sending side it starts up ten delivries at once. Some
> of them return with the "communication_with_mail_server_failed" message. On
> the receiving side I see that some of them end with status 28416 as above.
> Some of them simply end with status 0 however I don't get an OK messages on
> the receiving side (e.g. ...ok 1075813267 qp 23974 ddc saved 43 percent) for
> all of the ending connections with status 0.
> 
> Let me know what could be the cause of this problem or how I should analyze
> the log-files. Thanks.
> 

Hmm. I will try to find out why qmail-qmqpd blocks after printing the OK
line. Seems very strange.

-- 
:wq Claudio

Reply via email to