Re: peculiar lmtpd behavior

2002-12-23 Thread Henrique de Moraes Holschuh
Frobbing with the lmtp code, I noticed this thread...

On Sat, 22 Sep 2001, Kari Hurtta wrote:
  We believe it's this bit of code from savemsg() in lmtpengine.c:
  
  if (!strchr(rpath, '@')) {
  hostname = config_servername;
  }
  fprintf(f, Return-Path: %s%s%s\r\n,
  rpath, hostname ? @ : , hostname ? hostname : 
  
  If anyone can help us understand the reasons for this, we'd 
  greatly appreciate it.  
 
 Specially that does not take account
 Return-Path: 
 which is correct return batch for bounces.

Well, for an empty return path, rpath is NULL, so the code never gets
called.  If it is NOT NULL, but still empty, then there is a nasty bug
somewhere else.

-- 
  One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie. -- The Silicon Valley Tarot
  Henrique Holschuh



Re: peculiar lmtpd behavior

2001-09-22 Thread Kari Hurtta

 In our configuration the set of users that can receive mail locally is
 the contents of the assign file.
 Messages are passed to deliver via the users' .qmail files.
 
 The transformation of the Return-Path happens after qmail has
 handed off the message to cyrus for final delivery.
 
 We believe it's this bit of code from savemsg() in lmtpengine.c:
 
 if (!strchr(rpath, '@')) {
 hostname = config_servername;
 }
 fprintf(f, Return-Path: %s%s%s\r\n,
 rpath, hostname ? @ : , hostname ? hostname : 
 
 If anyone can help us understand the reasons for this, we'd 
 greatly appreciate it.  


Looks buggy :-)

Specially that does not take account

Return-Path: 

which is correct return batch for bounces.

-- 
  /\   |  Kari 
  \ / ASCII Ribbon Campaign |Hurtta
   X  Against HTML Mail |
  / \   |



Re: peculiar lmtpd behavior

2001-09-22 Thread Pascal Gienger

In our configuration the set of users that can receive mail locally is
the contents of the assign file.
Messages are passed to deliver via the users' .qmail files.

The transformation of the Return-Path happens after qmail has
handed off the message to cyrus for final delivery.

We believe it's this bit of code from savemsg() in lmtpengine.c:

 if (!strchr(rpath, '@')) {
 hostname = config_servername;
 }
 fprintf(f, Return-Path: %s%s%s\r\n,
 rpath, hostname ? @ : , hostname ? hostname : 

If anyone can help us understand the reasons for this, we'd
greatly appreciate it.


Ah NOW I understand. You mean the final delivery of a bounce generated by
any other MTA to the sender of the erroneous message (the sender has 
a cyrus imap mailbox). Yes then you are right, I think this is a bug.

There should be some matching for  as Return-Path in lmtpengine.c, because
this is the only correct Return-Path for an automatically generated bounce
message... :(

Pascal
-- 



Re: peculiar lmtpd behavior

2001-09-21 Thread Pascal Gienger

We are experiencing a behavior in lmtpd such that it inserts
an @ sign and hostname in the Return-Path if the Return-Path
does not contain an @ sign. So bounces on our system have a
Return-Path with @hostname and no username.

(As an aside, we are running qmail as opposed to sendmail.)

We are investigating this further and would appreciate
any assistance in understanding this behavior and
identifying a resolution.

As a first hint, I think you must search at qmail's configuration because
Cyrus cannot produce a bounce, can it?

Cyrus only understands LMTP and IMAP and both of them do not generate bounces
(Sieve Scripts are a different story, because they are using
/usr/lib/sendmail when forwarding mail).

It would be nice if you could tell us your configuration (how did you
couple QMAIL with Cyrus? via /var/users/assign?) and then we can look if we
can help you (sorry for the we as I am very new to this mailing list but
very old when it goes to Cyrus IMAPD things, I am using this beast from Ver-
sions 1.5 on and now we have 2.0.16 working like a charm).

Pascal Gienger
Finesse Networks, Germany
-- 



Re: peculiar lmtpd behavior

2001-09-21 Thread Tarjei Huse


 Cyrus only understands LMTP and IMAP and both of them do not generate
 bounces (Sieve Scripts are a different story, because they are using
 /usr/lib/sendmail when forwarding mail).
Where can I change that to using /usr/sbin/sendmail ?

:) Tarjei






Re: peculiar lmtpd behavior

2001-09-21 Thread Ken Murchison



Tarjei Huse wrote:
 
  Cyrus only understands LMTP and IMAP and both of them do not generate
  bounces (Sieve Scripts are a different story, because they are using
  /usr/lib/sendmail when forwarding mail).
 Where can I change that to using /usr/sbin/sendmail ?

Read imapd.conf(5).

Ken
-- 
Kenneth Murchison Oceana Matrix Ltd.
Software Engineer 21 Princeton Place
716-662-8973 x26  Orchard Park, NY 14127
--PGP Public Key--http://www.oceana.com/~ken/ksm.pgp



Re: peculiar lmtpd behavior

2001-09-21 Thread Lisa Maira

In our configuration the set of users that can receive mail locally is
the contents of the assign file.
Messages are passed to deliver via the users' .qmail files.

The transformation of the Return-Path happens after qmail has
handed off the message to cyrus for final delivery.

We believe it's this bit of code from savemsg() in lmtpengine.c:

if (!strchr(rpath, '@')) {
hostname = config_servername;
}
fprintf(f, Return-Path: %s%s%s\r\n,
rpath, hostname ? @ : , hostname ? hostname : 

If anyone can help us understand the reasons for this, we'd 
greatly appreciate it.  

thanks
Lisa Maira

On Fri, Sep 21, 2001 at 09:40:44AM, Pascal Gienger [EMAIL PROTECTED] wrote:
 
 It would be nice if you could tell us your configuration (how did you
 couple QMAIL with Cyrus? via /var/users/assign?) and then we can look if we
 can help you (sorry for the we as I am very new to this mailing list but
 very old when it goes to Cyrus IMAPD things, I am using this beast from Ver-
 sions 1.5 on and now we have 2.0.16 working like a charm).