Re: Zmailer to Exchange bug ?

[Mark Visser <mark@dnd.utwente.nl>]

On Mon, 20 May 2002, Matti Aarnio wrote:

> On Sun, May 19, 2002 at 11:33:17PM +0200, Mark Visser wrote:
> > Hm,
> >
> > Lately we noticed on our boxes that we got hanging smtpservers.
> > The version we're running right now is: ZMailer Server 2.99.55
> >
> > What seems to happen is that after getting to BDAT state the connections
> > hang and Zmailer nor the remote end closes them, causing smtpserver to
> > just say sleep, sleep sleep (according to strace)
> >
> > The boxes to which we see hanging smtpservers all have similar versions:
> >
> > Microsoft ESMTP MAIL Service, Version: 5.0.2195.4453 ready at..
> > Microsoft ESMTP MAIL Service, Version: 5.0.2195.4905 ready at..
> > Microsoft ESMTP MAIL Service, Version: 5.0.2195.2966 ready at..
> >
> > A sample one from ps says:
> >
> > daemon    1411  0.0  0.3 1960 1504 ?        S   May 13  0:00
> > <tacv-internet.tacv.com BDAT 129671 LAST net.tacv.com BDAT 129671 LAST
> > .tacv.com BDAT 129671 LAST .com BDAT 129671 LAST m BDAT 129671 LAST
> >
> > # truss -p 1411
> >     *** SUID: ruid/euid/suid = 0 / 1 / 1  ***
> > read(0, 0xEFFFD310, 8192)       (sleeping...)
> >
> > and that's all there is.
>
>    I have been rather confused with this report, and apparently Arnt
>    was too, as he referred to the outgoing smtp TA feature disabling..

My apogolies for the confusion, *note to self: first read over what you
typed, even if it's late*

As was hopefully clear a little later on, we're not talking about
smptserver, but about smtp process, the sending out from queue to remote.

>    For incoming  smtpserver's feature control you can use PARAM
>    statements at   $MAILSHARE/smtpserver.conf  file.  Specifically
>
>        PARAM NoCHUNKING
>
>    to disable BDAT support in reception.

I'll give that a try yes. It's 4 boxes, 2 of each kind doing it, so
testing with one of them should easily show diffs. They are identical for
the rest.

> > Did people see this before ? I tried 2.99.56 snapshot some time ago, but
> > that kept mail instead of sending it on (i didn't see quickly why, so i
> > backed it out).
>
>    Some people have asked me to make a release ASAP..  Yes, I should.
>    It has been about a year since previous one, and I have been hacking
>    in the CVS since then (not having very much time, though..)
>
>    11 months ago (after 55-patch1) I addressed some of timeout problem
>    issues at the  smtpserver,  and now it looks like I am not seeing
>    any stuck servers.  (That might solve the thing you see.)

Ok, i grabbed the CVS copy of today and compiled it.
Few warnings but compiles with less warnings then the previous one. If you
want to have the output of configure & make (SunOS 5.6 SPARC) send me a
mail and i'll pass it on, but  if i remember well you had 2.6 boxes
yourself.

>    I am running the lattest code at several machines, and the only
>    excuse for not making a release is that I see some mighty strange
>    protocol errors from various places in general direction of eastern
>    Asia/China..  As if the smtp TA client code is somehow broken, and
>    is sending inappropriately formatted data out -- e.g. in places
>    where there should be  CRLF,  it is (claimed to be) sending an LF
>    alone..

When i tried 2.99.56-pre1 snapshot from CVS in january i noticed on our
etrn boxes that it kept all mail, as if it took 'queueonly' very literal.
(note..this is with the box getting ETRN requests).

I'll give the new CVS copy a try.

> > This happens on both ETRN as fallback MX boxes.
> > Any hints appreciated.
> > Mark
> --

Thanx Arnt & Matti !

Mark


-
To unsubscribe from this list: send the line "unsubscribe zmailer" in
the body of a message to majordomo@nic.funet.fi