qmail Digest 20 Jul 1999 10:00:01 -0000 Issue 703

Topics (messages 27932 through 27976):

ETRN
        27932 by: Frank Greven <[EMAIL PROTECTED]>
        27936 by: Ray Marshall <[EMAIL PROTECTED]>
        27937 by: Anand Buddhdev <[EMAIL PROTECTED]>
        27967 by: "Simon Elder" <[EMAIL PROTECTED]>
        27972 by: "Simon Elder" <[EMAIL PROTECTED]>

qmail analog
        27933 by: "dandiu" <[EMAIL PROTECTED]>

getting mail from my ISP with qmail
        27934 by: Yan Seiner <[EMAIL PROTECTED]>

Mass migration off of qmail because of lack of DSNs?
        27935 by: Russell Nelson <[EMAIL PROTECTED]>

qmail does not deliver certain messages and does not return them
        27938 by: "Holger van Koll" <[EMAIL PROTECTED]>
        27939 by: "Petr Novotny" <[EMAIL PROTECTED]>
        27943 by: "Holger van Koll" <[EMAIL PROTECTED]>
        27945 by: "Petr Novotny" <[EMAIL PROTECTED]>

ISP: Industry name for various email services?
        27940 by: Dave Kitabjian <[EMAIL PROTECTED]>

Virtual users
        27941 by: Tony Wade <[EMAIL PROTECTED]>
        27942 by: "Petr Novotny" <[EMAIL PROTECTED]>

What happened to email.com's clue?
        27944 by: Russell Nelson <[EMAIL PROTECTED]>
        27946 by: "Alex Miller" <[EMAIL PROTECTED]>

Advantages with qmail and using reiserfs???
        27947 by: Dave Sill <[EMAIL PROTECTED]>
        27950 by: Scott Schwartz <[EMAIL PROTECTED]>
        27952 by: Jason Haar <[EMAIL PROTECTED]>

Eliding quotes in envelope? (Re: New spammer exploit ...)
        27948 by: Bruno Wolff III <[EMAIL PROTECTED]>
        27949 by: Scott Schwartz <[EMAIL PROTECTED]>
        27953 by: Bruno Wolff III <[EMAIL PROTECTED]>
        27954 by: Scott Schwartz <[EMAIL PROTECTED]>
        27955 by: "Sam" <[EMAIL PROTECTED]>
        27957 by: Bruno Wolff III <[EMAIL PROTECTED]>
        27958 by: "Sam" <[EMAIL PROTECTED]>
        27959 by: Bruno Wolff III <[EMAIL PROTECTED]>
        27961 by: "Sam" <[EMAIL PROTECTED]>
        27962 by: Scott Schwartz <[EMAIL PROTECTED]>
        27963 by: Bruno Wolff III <[EMAIL PROTECTED]>
        27965 by: Bruno Wolff III <[EMAIL PROTECTED]>
        27968 by: Scott Schwartz <[EMAIL PROTECTED]>
        27973 by: Bruno Wolff III <[EMAIL PROTECTED]>

Logging tcpserver in.pop3d and cyclog
        27951 by: "Tim Hunter" <[EMAIL PROTECTED]>
        27971 by: Chris Johnson <[EMAIL PROTECTED]>

Compile Qmail Staticly
        27956 by: "Robert J. Adams" <[EMAIL PROTECTED]>

Login characters limited?
        27960 by: Sienna <[EMAIL PROTECTED]>
        27976 by: Anand Buddhdev <[EMAIL PROTECTED]>

Return-Path
        27964 by: David Villeger <[EMAIL PROTECTED]>
        27966 by: Bruno Wolff III <[EMAIL PROTECTED]>
        27969 by: "Sam" <[EMAIL PROTECTED]>
        27970 by: Russell Nelson <[EMAIL PROTECTED]>

basic question
        27974 by: Quinn Coldiron <[EMAIL PROTECTED]>
        27975 by: Anand Buddhdev <[EMAIL PROTECTED]>

Administrivia:

To subscribe to the digest, e-mail:
        [EMAIL PROTECTED]

To unsubscribe from the digest, e-mail:
        [EMAIL PROTECTED]

To bug my human owner, e-mail:
        [EMAIL PROTECTED]

To post to the list, e-mail:
        [EMAIL PROTECTED]


----------------------------------------------------------------------


He,

I'am running qmail 1.03 on an RH 6.0 (2.2.5) box and I host mail for special
domain. 
Our client's mail server (no Linux/Unix maschine) makes a dial-up connection to
the internet and then want's to to get his emails delivered from our server.

I've heard something about ETRN which is applied but I don't know what exactly
to do.

Can anybody help me out?

Thanks a lot,
Frank

/---------------------------------------------------------\
|  RIMNet GmbH                  Tel. :  (02102) 420 760   |
|  Kaiserswerther Str. 115      Fax  :  (02102) 420 62    |
|  D-40880 Ratingen             WWW  :  www.rimnet.de     |
|  Germany                      Email:  [EMAIL PROTECTED]    |




Frank,

On Mon, 19 Jul 1999, you wrote:

> I'am running qmail 1.03 on an RH 6.0 (2.2.5) box and I host mail for
> special domain.  Our client's mail server (no Linux/Unix maschine) makes
> a dial-up connection to the internet and then want's to to get his
> emails delivered from our server.
>
> I've heard something about ETRN which is applied but I don't know what
> exactly to do.
>
> Can anybody help me out?

I don't know where, or if this is documented.  Here's what I know from
personal experience.


ETRN is an SMTP command, issued by a client machine, to it's SMTP server,
to tell that server to send along any SMTP Email queued up for it.

The sendmail package has a contributed utility called etrn.pl which issues
the appropriate command.  There is also a script file called etrn.sh to do
the same thing.  These script files use to be how people had to pickup
their SMTP mail.

But, later versions of fetchmail (I'm using 4.6) have the ability to issue
that command.  So the script files are no longer needed.

I suggest that your client just add to the end of the file .fetchmailrc ,
a line like the following:

  poll <SMTP server> with protocol ETRN

For example, if the SMTP server's name is smtp.rimnet.de , the line would
look like this:

  poll smtp.rimnet.de with protocol ETRN

For multiple SMTP servers, multiple lines can be used.

If you are serving POP mail, then your client should already be using
sendmail, and already have .fetchmailrc in the correct place(s).  If not,
it must live in the home directory, of the account from which fetchmail
will be executed.

I execute it from /etc/ppp/ip-up.local , so I have the file in root's
home directory.

This poll command only has to be issued once per dialing connection -- the
server will continue sending newly received Email, as long as the client
remains connected to the net.

/  Ray
------------------------------------+------------------------------------
Ray Marshall <[EMAIL PROTECTED]>        | Unconditional Forgiveness & Love --
Chapel Hill NC or Sutton Mills NH   |   The corner stones of coexistence.
Work: [EMAIL PROTECTED]





On Mon, Jul 19, 1999 at 02:10:16PM +0200, Frank Greven wrote:

Ray has provided a solution for the client end, ie. to use fetchmail.
However, on your end, you need to setup a catch-all Maildir for them.
Detailed instructions for this are available in the serialmail package,
which you will need. download:

ftp://koobera.math.uic.edu/pub/software/serialmail-0.75.tar.gz

read the instructions and you'll be fine.

> He,
> 
> I'am running qmail 1.03 on an RH 6.0 (2.2.5) box and I host mail for special
> domain. 
> Our client's mail server (no Linux/Unix maschine) makes a dial-up connection to
> the internet and then want's to to get his emails delivered from our server.
> 
> I've heard something about ETRN which is applied but I don't know what exactly
> to do.

-- 
See complete headers for more info




Another way I have found to do this is to write a script that runs from a
cron job say ~ every 10 minutes that checks if the customer has dialled into
you term server. If they have then the script just does a
'killall -HUP qmail-send' .. which forces everything deferred in the queue
to be sent again.

Not a pretty solution but it works for my small queue.


-----Original Message-----
From: Frank Greven <[EMAIL PROTECTED]>
To: [EMAIL PROTECTED] <[EMAIL PROTECTED]>
Date: Monday, 19 July 1999 10:11
Subject: ETRN


>He,
>
>I'am running qmail 1.03 on an RH 6.0 (2.2.5) box and I host mail for
special
>domain.
>Our client's mail server (no Linux/Unix maschine) makes a dial-up
connection to
>the internet and then want's to to get his emails delivered from our
server.
>
>I've heard something about ETRN which is applied but I don't know what
exactly
>to do.
>
>Can anybody help me out?
>
>Thanks a lot,
>Frank
>
>/---------------------------------------------------------\
>|  RIMNet GmbH                  Tel. :  (02102) 420 760   |
>|  Kaiserswerther Str. 115      Fax  :  (02102) 420 62    |
>|  D-40880 Ratingen             WWW  :  www.rimnet.de     |
>|  Germany                      Email:  [EMAIL PROTECTED]    |
>





Woops that should be killall -ALRM qmail-send

-----Original Message-----
From: Simon Elder <[EMAIL PROTECTED]>
To: [EMAIL PROTECTED] <[EMAIL PROTECTED]>
Date: Tuesday, 20 July 1999 9:37
Subject: Re: ETRN


>Another way I have found to do this is to write a script that runs from a
>cron job say ~ every 10 minutes that checks if the customer has dialled
into
>you term server. If they have then the script just does a
>'killall -HUP qmail-send' .. which forces everything deferred in the queue
>to be sent again.
>
>Not a pretty solution but it works for my small queue.
>
>
>-----Original Message-----
>From: Frank Greven <[EMAIL PROTECTED]>
>To: [EMAIL PROTECTED] <[EMAIL PROTECTED]>
>Date: Monday, 19 July 1999 10:11
>Subject: ETRN
>
>
>>He,
>>
>>I'am running qmail 1.03 on an RH 6.0 (2.2.5) box and I host mail for
>special
>>domain.
>>Our client's mail server (no Linux/Unix maschine) makes a dial-up
>connection to
>>the internet and then want's to to get his emails delivered from our
>server.
>>
>>I've heard something about ETRN which is applied but I don't know what
>exactly
>>to do.
>>
>>Can anybody help me out?
>>
>>Thanks a lot,
>>Frank
>>
>>/---------------------------------------------------------\
>>|  RIMNet GmbH                  Tel. :  (02102) 420 760   |
>>|  Kaiserswerther Str. 115      Fax  :  (02102) 420 62    |
>>|  D-40880 Ratingen             WWW  :  www.rimnet.de     |
>>|  Germany                      Email:  [EMAIL PROTECTED]    |
>>
>
>





Can anyone please tell me what outputs the
"x-commands" will generate & the syntax for utilizing 
"x-commands" (i.e. xsenders/xqp/xrec.) within QMAIL Analog?  
I have gotten the "z-commands" to give certain outputs

Newbie to QMAIL

Thnx in ADV

Ume






I collect about 2 doz. accts from 2 (soon to be 3)domains.  I was forced
to set up a dummy user (mailrelay) that fetchmail delivers all mail to. 
This account has a .procmailrc which then sorts the mail and retransmits
to the appropriate local user.

Yan

Sim wrote:
> 
> On Fri, 16 Jul 1999, Dave Sill wrote:
> > [EMAIL PROTECTED] wrote:
> > >
> > >You'll need fetchmail to download the mails and pass them to qmail fo
> > >rlocal delivery.  However, this is a very difficult thing to get set
> > >up - I did have it working once, but now its broken and all incoming
> > >mail is not recognised and passed to postmaster
> >
> > For a sample .fetchmailrc, see:
> >
> >    http://Web.InfoAve.Net/~dsill/lwq.html#fetchmail
> 
> This is more or less what I have, but it does not achieve my goal.  I have a
> single mailbox, M, from my ISP with two aliases X and Y (in the form of
> [EMAIL PROTECTED]) and I want these to be delivered to x and y
> (shortenings of our first names) on my linux box.  Y's mails are always
> adressed only to Y, but X receives mails from various listservs etc.
> 
> I thought the way to do it was to run fetchmail as deamon from rc.local with
> /etc/.fetchmailrc
> 
> poll mail.tvd.be proto pop3 nodns
> user M with password P is * here
> fetchall forcecr keep to * here
> 
> and have a .qmail-X pointing to x and .qmail-Y pointing to y in ~alias.  But
> this does not deliver the results.  What would you recommend?
> 
> Simon

-- 

           __      __
          | /      /
           /------/
       -- / \    / \ --
     /   /\  \  /  /\   \
    |   /  |  \/--|--    |
     \    /        \    /
       ~~            ~~

"The older I get, the faster I was."




Vince Vielhaber writes:
 > 
 > This thread died two months ago.

 > > > > Huh?  What's your threshold for "widely supported"?  Doesn't sendmail
 > > > > have something like 80% market share and nice DSN support?

Not to mention the fact that sendmail's market share is 63% and slipping.

-- 
-russ nelson <[EMAIL PROTECTED]>  http://crynwr.com/~nelson
Crynwr supports Open Source(tm) Software| PGPok | Government schools are so
521 Pleasant Valley Rd. | +1 315 268 1925 voice | bad that any rank amateur
Potsdam, NY 13676-3213  | +1 315 268 9201 FAX   | can outdo them. Homeschool!




Problem:

Sometimes I see a line like that in /var/log/maillog

Jul 19 14:20:17 kserver qmail: 932386817.166513 starting delivery 8: msg
1331273 to local @kserver.localdomain
Jul 19 14:20:17 kserver qmail: 932386817.169277 delivery 8: success:

Last time that happened it was because the reply-to Adress of the sender was
something like "holger"@ <vankoll.de> instead of [EMAIL PROTECTED]

Look at this:
[root@kserver log]# telnet localhost 25
Trying 127.0.0.1...
Connected to localhost.
Escape character is '^]'.
220 kserver.localdomain ESMTP
helo sdf
250 kserver.localdomain
mail <[EMAIL PROTECTED]>
250 ok
rcpt "holger" <@vankoll.de>
250 ok
data
354 go ahead
this is
a test
.
250 ok 932391185 qp 7414

Jul 19 15:33:05 kserver qmail: 932391185.868144 info msg 1331273: bytes 208
from <[EMAIL PROTECTED]> qp 7414 uid 0
Jul 19 15:33:05 kserver qmail: 932391185.871202 starting delivery 38: msg
1331273 to local @kserver.localdomain
Jul 19 15:33:05 kserver qmail: 932391185.872089 status: local 1/10 remote
0/20
Jul 19 15:33:05 kserver qmail: 932391185.874862 delivery 38: success:
Jul 19 15:33:05 kserver qmail: 932391185.875912 status: local 0/10 remote
0/20
Jul 19 15:33:05 kserver qmail: 932391185.876913 end msg 1331273


Qmail states that everything is fine, but that message will never be
received.

I know that the e-mail address in the example is wrong, but IMHO it is wrong
behaviour NOT to deliver and NOT to bounce the email.

Unfortunately, "holger" <@vankoll.de> is not the only way to make qmail
behave like that.

I also noticed that using mailto: within a webpage can create such
"deliveries" ; especially when using netscape as browser.

Or here is another example:

telnet localhost 25
Trying 127.0.0.1...
Connected to localhost.
Escape character is '^]'.
220 kserver.localdomain ESMTP
helo asdf
250 kserver.localdomain
mail "[EMAIL PROTECTED]"
250 ok
rcpt "[EMAIL PROTECTED]"
250 ok
data
354 go ahead

Subject: Test
.
250 ok 932391629 qp 7540


Jul 19 15:40:28 kserver qmail: 932391628.673799 new msg 1331273
Jul 19 15:40:28 kserver qmail: 932391628.674163 info msg 1331273: bytes 209
from <> qp 7540 uid 0
Jul 19 15:40:28 kserver qmail: 932391628.867030 starting delivery 43: msg
1331273 to local @kserver.localdomain
Jul 19 15:40:28 kserver qmail: 932391628.867218 status: local 1/10 remote
0/20
Jul 19 15:40:28 kserver qmail: 932391628.966109 delivery 43: success:
Jul 19 15:40:28 kserver qmail: 932391628.967150 status: local 0/10 remote
0/20
Jul 19 15:40:28 kserver qmail: 932391628.967635 end msg 1331273



It is too difficult for me to "fix" qmail regarding this. But is there -any-
way to catch those messages?







-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

[snip]
> Qmail states that everything is fine, but that message will never be
> received.
> 
> I know that the e-mail address in the example is wrong, but IMHO it is
> wrong behaviour NOT to deliver and NOT to bounce the email.
> 
> Unfortunately, "holger" <@vankoll.de> is not the only way to make qmail
> behave like that.

I fail to see what you consider erratic behaviour. Qmail received a 
message and successfully delivered it to zero recipients. Where 
should a bounce come from? (Your RCPT TO has empty local part -
 therefore zero recipients.)

If it's an error, it's an error on the sending side. Fix that one.

Qmail does just good.

-----BEGIN PGP SIGNATURE-----
Version: PGP 6.0.2 -- QDPGP 2.60 
Comment: http://community.wow.net/grt/qdpgp.html

iQA/AwUBN5M7dVMwP8g7qbw/EQIK9QCfYn3w36KwVxNXUlkfBMBBaTLT2DEAnRxU
EQDcojP5Q1CpUQL4J8cX7WoY
=vvJz
-----END PGP SIGNATURE-----
--
Petr Novotny, ANTEK CS
[EMAIL PROTECTED]
http://www.antek.cz
PGP key ID: 0x3BA9BC3F
-- Don't you know there ain't no devil there's just God when he's drunk.
                                                             [Tom Waits]




>> Unfortunately, "holger" <@vankoll.de> is not the only way to make qmail
>> behave like that.

>I fail to see what you consider erratic behaviour. Qmail received a
>message and successfully delivered it to zero recipients.

Where is the sense in delivering to zero recipients?
Can this be called delivering?

>Where
>should a bounce come from? (Your RCPT TO has empty local part -
>therefore zero recipients.)


IMHO it COULD bounce that message. I cant tell that it must because I dont
know the concerning specs.

>If it's an error, it's an error on the sending side. Fix that one.

I agree that there is an error on the sending side. But how can the user
find out that? By checking the email address of the recipient every time he
replies to a message?

>Qmail does just good.
Surely it does. But it could do a bit better in that case...






-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

> >I fail to see what you consider erratic behaviour. Qmail received a
> >message and successfully delivered it to zero recipients.
> 
> Where is the sense in delivering to zero recipients?
> Can this be called delivering?

Well, unless I'm mistaken, ezmlm-idx relies on this behaviour - for 
example it can send digests to zero recipients (noone is 
subscribed to digest version). If you change qmail's behaviour by a 
patch, you mey get problems with ezmlm_idx.

> IMHO it COULD bounce that message. I cant tell that it must because I dont
> know the concerning specs.

I tried to search the specs but couldn't find an authoritative 
resolution. By bouncing messages with corrent SMTP headers, 
you're becoming just-one-more-broken-SMTP-client. No big deal 
though.

> I agree that there is an error on the sending side. But how can the user
> find out that? By checking the email address of the recipient every time
> he replies to a message?

If you think that MTA should correct errors made by silly users and 
broken MUAs - well, that's sendmailism. sendmail might just suit 
your needs better.

-----BEGIN PGP SIGNATURE-----
Version: PGP 6.0.2 -- QDPGP 2.60 
Comment: http://community.wow.net/grt/qdpgp.html

iQA/AwUBN5NDWFMwP8g7qbw/EQIcrACfY3jW/qiEo4hVJP9KG5VJvzGigpgAnRs2
4USyBYgEc9bamQhZ7M1hSMjm
=n8pp
-----END PGP SIGNATURE-----
--
Petr Novotny, ANTEK CS
[EMAIL PROTECTED]
http://www.antek.cz
PGP key ID: 0x3BA9BC3F
-- Don't you know there ain't no devil there's just God when he's drunk.
                                                             [Tom Waits]





What are the industry-standard names for the following email services:

------------------------------------------------------------------------  
-----
1) They can have:

        <any_name>@<their_domain>

and a separate POP for each. (.qmail-<any_name>)
------------------------------------------------------------------------  
-----
2) They can have:

        <any_name>@<their_domain>

but they get only a single pop account, "default", for retrieval of all the 
mail (.qmail-default).
------------------------------------------------------------------------  
-----
3) All mail to

        <any_name>@<their_domain_2>

is treated as

        <any_name>@<their_domain_1>

(by virtualdomains entry: <their_domain_2>:<their_domain_1>)
------------------------------------------------------------------------  
-----

We've been calling these variously "Full Domain email", "POP domain email", 
"Alias Domain email", etc. But I want to know what the *real* names are for 
these. Thanks very much!

Dave





Hi all, 

I was wondering how the following would work. 

I have a few domains which o am going to run as virtualdomain and others
which are just going to have a smtproute to another server. I want to have
some Virtual users deliver to another machine and the remainder of the users
deliver to yet another server. 

ie. 

postmaster@domain deliver to server1
root@domain deliver to server1
owner@domain deliver to server1

anything else that is for that domain deliver to server2

can this be done ? 


Tony Wade
The Internet Solution
Tel:    (+27 11) 283 5483
Fax:    (+27 11) 283 5401
E-mail: [EMAIL PROTECTED] 
Web:    http://www.is.co.za
#include <std/disclaimer.h>






-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

> I was wondering how the following would work. 
> 
> I have a few domains which o am going to run as virtualdomain and others
> which are just going to have a smtproute to another server. I want to have
> some Virtual users deliver to another machine and the remainder of the
> users deliver to yet another server. 
> 
> ie. 
> 
> postmaster@domain deliver to server1
> root@domain deliver to server1
> owner@domain deliver to server1
> 
> anything else that is for that domain deliver to server2
> 
> can this be done ? 

I would do that without virtual domains at all.

1. Put domain into locals.
(2. If you have local users, make sure that you use users/assign 
mechanism to override them. Routing everything to alias is just 
fine.)
3. Arrange local delivery for postmaster, owner, root (for example 
by ~alias/.qmail-root etc.)
4. In ~alias/.qmail-default, do
|forward "$LOCAL"@server2
(5. You may need to arrange delivery to server2 via smtproutes.)

-----BEGIN PGP SIGNATURE-----
Version: PGP 6.0.2 -- QDPGP 2.60 
Comment: http://community.wow.net/grt/qdpgp.html

iQA/AwUBN5NAe1MwP8g7qbw/EQISigCfX0G3R9HQkSFd3hTflQd1uDBbBeEAn0Ef
5kw+qAMJNZpnEKsoczOPrEpO
=EHIX
-----END PGP SIGNATURE-----
--
Petr Novotny, ANTEK CS
[EMAIL PROTECTED]
http://www.antek.cz
PGP key ID: 0x3BA9BC3F
-- Don't you know there ain't no devil there's just God when he's drunk.
                                                             [Tom Waits]




What happened to email.com & mail.com's clue?  They're munging
envelope sender addresses by deleting everything up to an equal sign.
This is non-RFC behavior!  You can see what I mean by telnetting to
port 25 on an email.com MX and issuing the following smtp commands.
I'm presuming that your username is USER and your host is HOST.

helo HOST
rcpt from: <deletethisjunk=USER@HOST>
mail to: <[EMAIL PROTECTED]>
data
From: me
To: you
.

In spite of your email address NOT being "deletethisjunk=USER@HOST",
you will receive the bounce message!

This affects ezmlm admins because ezmlm requires on the equal sign
being respected.  When a similar problem came up at a customer's site,
I change from using the single character '=' to using the
two-character pair '.-'.  Domain names cannot begin with a dash, so
there is no possible ambiguity.  And many email addresses and domain
names have dots and dashes in them, so there is less lossage
associated with '.-'.  Sigh.

-- 
-russ nelson <[EMAIL PROTECTED]>  http://crynwr.com/~nelson
Crynwr supports Open Source(tm) Software| PGPok | Government schools are so
521 Pleasant Valley Rd. | +1 315 268 1925 voice | bad that any rank amateur
Potsdam, NY 13676-3213  | +1 315 268 9201 FAX   | can outdo them. Homeschool!




Is yahoo.com's email owned (and operated) by
mail.com or email.com?

I've been having a subscription failure
from yahoo.com

Perhaps they are doing the same silly thing.

However, if you convert to using ".-" that will
completely change the way subscriptions already
work. Everyone who has a set of instructions
will be expecting the "=" sign to work.

Maybe aol.com and mail.com are subtly trying
to create some havoc, so the community of standards
will be replaced by corporate standards.

Alex Miller

> -----Original Message-----
> From: Russell Nelson [mailto:[EMAIL PROTECTED]]
> Sent: Monday, July 19, 1999 10:18 AM
> To: [EMAIL PROTECTED]
> Subject: What happened to email.com's clue?
> 
> 
> What happened to email.com & mail.com's clue?  They're munging
> envelope sender addresses by deleting everything up to an equal sign.
> This is non-RFC behavior!  You can see what I mean by telnetting to
> port 25 on an email.com MX and issuing the following smtp commands.
> I'm presuming that your username is USER and your host is HOST.
> 
> helo HOST
> rcpt from: <deletethisjunk=USER@HOST>
> mail to: <[EMAIL PROTECTED]>
> data
> From: me
> To: you
> .
> 
> In spite of your email address NOT being "deletethisjunk=USER@HOST",
> you will receive the bounce message!
> 
> This affects ezmlm admins because ezmlm requires on the equal sign
> being respected.  When a similar problem came up at a customer's site,
> I change from using the single character '=' to using the
> two-character pair '.-'.  Domain names cannot begin with a dash, so
> there is no possible ambiguity.  And many email addresses and domain
> names have dots and dashes in them, so there is less lossage
> associated with '.-'.  Sigh.
> 
> -- 
> -russ nelson <[EMAIL PROTECTED]>  http://crynwr.com/~nelson
> Crynwr supports Open Source(tm) Software| PGPok | Government 
> schools are so
> 521 Pleasant Valley Rd. | +1 315 268 1925 voice | bad that any 
> rank amateur
> Potsdam, NY 13676-3213  | +1 315 268 9201 FAX   | can outdo them. 
> Homeschool!
> 




Troy Morrison <[EMAIL PROTECTED]> wrote:
>
>We have a fairly ongoing problem with some of the users at work who don't
>seem capable of cleaning out their INBOX, so they end up with 100MB mail
>spools with 7000 messages in them.
>
>I had theorized that chunking over the 100MB mailbox was slow, and that
>using a maildir would be much faster, and it is except that with that many
>messages, the OS slows down.
>
>Does anyone have any suggestions (including a different filesystem -- we
>just used ext2) that might help this?

A hashing filesystem, such as SGI's XFS (which is coming to Linux),
makes large directories much more manageable. But even with zero
filesystem overhead, most MUA's will only handle mailboxes so big with
acceptable performance: they have to open each message file and
extract various header information. A clever MUA could maintain that
information separately, but none that I'm aware of actually do.

-Dave




Dave Sill <[EMAIL PROTECTED]> writes:
| A clever MUA could maintain that
| information separately, but none that I'm aware of actually do.

What about xmh and exmh?  (I don't think they properly mutex the cache,
though.)





> most MUA's will only handle mailboxes so big with
> acceptable performance: they have to open each message file and
> extract various header information. A clever MUA could maintain that
> information separately, but none that I'm aware of actually do.

I think the Cyrus IMAP server falls into this category. It keeps a "meta file" 
of header data for IMAP header calls...

-- 
Cheers

Jason Haar

Unix/Network Specialist, Trimble NZ
Phone: +64 3 3391 377 Fax: +64 3 3391 417
     





On Mon, Jul 19, 1999 at 01:25:15AM -0400,
  Scott Schwartz <[EMAIL PROTECTED]> wrote:
> Russell Nelson <[EMAIL PROTECTED]> writes:
> | Command-line arguments are RFC821 addresses, but body addresses are
> | RFC822 addresses.
> 
> I'm only talking about the envelope (rfc821 addresses).  It's
> inconsistent that qmail-smtpd strips the quotes from the envelope while
> qmail-inject doesn't.

I ask this here a while back. The answer is that the arguments to qmail-inject
are raw email addresses, they are NOT encoded. It is annoying that sendmail
does things differently.

This can cause problems for programs, such as mutt, when using addresses
that need to be quoted, since the arguments it needs to pass are different
for qmail-inject and sendmail.




Bruno Wolff III <[EMAIL PROTECTED]> writes:
| I ask this here a while back. The answer is that the arguments to qmail-inject
| are raw email addresses, they are NOT encoded.

Shouldn't that also be true of the address in RCPT TO during an SMTP session?





On Mon, Jul 19, 1999 at 12:31:31PM -0400,
  Scott Schwartz <[EMAIL PROTECTED]> wrote:
> Bruno Wolff III <[EMAIL PROTECTED]> writes:
> | I ask this here a while back. The answer is that the arguments to qmail-inject
> | are raw email addresses, they are NOT encoded.
> 
> Shouldn't that also be true of the address in RCPT TO during an SMTP session?

No. That address is encoded according to rfc 821.

Addresses in the headers are encoded according to rfc 822.




Bruno Wolff III <[EMAIL PROTECTED]> writes:
| No. That address is encoded according to rfc 821.
| Addresses in the headers are encoded according to rfc 822.

I am not talking about rfc822 headers, only about rfc821 envelopes.

To reiterate:  qmail-smtpd strips quotes from RFC821 ENVELOPES (the
stuff you feed it with RCPT TO), and qmail-inject doesn't (the stuff
you feed it on the command line).  That's inconsistent, right?





Scott Schwartz writes:

> Bruno Wolff III <[EMAIL PROTECTED]> writes:
> | No. That address is encoded according to rfc 821.
> | Addresses in the headers are encoded according to rfc 822.
> 
> I am not talking about rfc822 headers, only about rfc821 envelopes.
> 
> To reiterate:  qmail-smtpd strips quotes from RFC821 ENVELOPES (the
> stuff you feed it with RCPT TO), and qmail-inject doesn't (the stuff
> you feed it on the command line).  That's inconsistent, right?

Right.

Over the last 7-8 months I have studied, religiously, RFC821, 822,
1891-1894, 2045, and several other E-mail related RFCs, as I was in the
process of writing what right now stands to be about 2.5 megabytes worth of
source code (a lot of that is actually GNU's autobloat, but I won't get
into that right now).

Unless the envelope address refers to a local mailbox, qmail-smtpd has no
business stripping quotes from the local part of the address, in most
cases.  It is arguable whether it should do that to the domain portion of
the address, but it really has no business doing that to the local portion.

There may be some point of argument if the quoted portion can be accurately
transcribed as an an RFC822 atom, in which case an argument can be made
that quotes can be stripped.  However, stripping quotes unilaterally is a
completely broken behavior.


-- 
Sam





On Mon, Jul 19, 1999 at 09:43:14PM +0000,
> 
> There may be some point of argument if the quoted portion can be accurately
> transcribed as an an RFC822 atom, in which case an argument can be made
> that quotes can be stripped.  However, stripping quotes unilaterally is a
> completely broken behavior.

I disagree. It is pretty clear that semanticly the quotes are not part
of the local address. This allows you to put characters in the local
part of the address that would conflict with other things allowed by
rfc 821, such as routing information and the like.

SMTP servers that treat "local"@example.com and [EMAIL PROTECTED] as
different are broken.




Bruno Wolff III writes:

> On Mon, Jul 19, 1999 at 09:43:14PM +0000,
> > 
> > There may be some point of argument if the quoted portion can be accurately
> > transcribed as an an RFC822 atom, in which case an argument can be made
> > that quotes can be stripped.  However, stripping quotes unilaterally is a
> > completely broken behavior.
> 
> I disagree. It is pretty clear that semanticly the quotes are not part
> of the local address. This allows you to put characters in the local
> part of the address that would conflict with other things allowed by
> rfc 821, such as routing information and the like.
> 
> SMTP servers that treat "local"@example.com and [EMAIL PROTECTED] as
> different are broken.

Only if the SMTP servers in question actually store the mailboxes for the
example.com domain, and even then that is still arguable.

Intermediate mails ervers really shouldn't mess around with things that do
not concern them.  Perhaps the local portion of the address contains
characters which may not be RFC822 tokens, but still may require quoting by
the recipient's server.

If you do not handle the mail for example.com, don't screw around with
example.com's local address.  Just pass it along to example.com, and let
them deal with it.

-- 
Sam





On Mon, Jul 19, 1999 at 10:04:47PM +0000,
  Sam <[EMAIL PROTECTED]> wrote:
> 
> If you do not handle the mail for example.com, don't screw around with
> example.com's local address.  Just pass it along to example.com, and let
> them deal with it.

The way that is handled is requoting this that need to be quoted.

For example if I relay mail to example.com I should be able to take
an address of ",a"."b"@example.com and convert it locally to the
real address of ,[EMAIL PROTECTED] and then requote it for the relay as
",a.b"@example.com and things should work just fine.
If you really want an address of "local"@example.com, it should be encoded
as "\"local\""@example.com.

If you look at the "Joe\,Smith" example in rfc 821, it speciifcally says
this represents a 9 character string with a comma being the fourth character.




Bruno Wolff III writes:

> On Mon, Jul 19, 1999 at 10:04:47PM +0000,
>   Sam <[EMAIL PROTECTED]> wrote:
> > 
> > If you do not handle the mail for example.com, don't screw around with
> > example.com's local address.  Just pass it along to example.com, and let
> > them deal with it.
> 
> The way that is handled is requoting this that need to be quoted.
> 
> For example if I relay mail to example.com I should be able to take
> an address of ",a"."b"@example.com and convert it locally to the
> real address of ,[EMAIL PROTECTED] and then requote it for the relay as
> ",a.b"@example.com and things should work just fine.

Perhaps, but that is completely unnecessary if you're going to relay it to
another server.

> If you really want an address of "local"@example.com, it should be encoded
> as "\"local\""@example.com.
> 
> If you look at the "Joe\,Smith" example in rfc 821, it speciifcally says
> this represents a 9 character string with a comma being the fourth character.

However, only the recipient's mail server cares about Joe,Smith. 
Intermediate mail servers don't care about it.

-- 
Sam





To make this more concrete, consider the following:

        RCPT TO:<"A B"@ARPA>

That can be parsed by the grammer given in rfc821.  On the other
hand, this cannot:

        RCPT TO:<A B@ARPA>

because ``A B'' contains an SP, which must be quoted somehow.

If you send a message via qmail-smtpd to the first recipient above, the
second one is what you'll see in the text of the bounce that qmail
sends you, when it mentions the original (sic) envelope address.





> > If you look at the "Joe\,Smith" example in rfc 821, it speciifcally says
> > this represents a 9 character string with a comma being the fourth character.
> 
> However, only the recipient's mail server cares about Joe,Smith. 
> Intermediate mail servers don't care about it.

Exactly. That is why any valid encoding of the same semantic information
should be just fine with the destination mail server.

If a server decides to just save the real address while deciding what
should be done with the message and then reencodes it (possibly differently
than originally) that should be just as valid as saving the original
encoding.




On Mon, Jul 19, 1999 at 06:30:51PM -0400,
  Scott Schwartz <[EMAIL PROTECTED]> wrote:
> To make this more concrete, consider the following:
> 
>       RCPT TO:<"A B"@ARPA>
> 
> That can be parsed by the grammer given in rfc821.  On the other
> hand, this cannot:
> 
>       RCPT TO:<A B@ARPA>
> 
> because ``A B'' contains an SP, which must be quoted somehow.
> 
> If you send a message via qmail-smtpd to the first recipient above, the
> second one is what you'll see in the text of the bounce that qmail
> sends you, when it mentions the original (sic) envelope address.

The original envelope address is:
A B@ARPA

It is only encoded as:
<"A B"@ARPA>




Bruno Wolff III <[EMAIL PROTECTED]> writes:
| The original envelope address is:
| A B@ARPA
| 
| It is only encoded as:
| <"A B"@ARPA>

Rather than arguing about what "is" means, I simply observe (again)
that for any value of "is" qmail-smtpd wrongly transforms <"A B"@ARPA>
into <A B@ARPA>, and wrongly behaves differently than qmail-inject.

It's not a question of encoding one thing two different, valid ways.
It's a question of transforming a valid encoding into an invalid one.





On Mon, Jul 19, 1999 at 08:03:45PM -0400,
  Scott Schwartz <[EMAIL PROTECTED]> wrote:
> Bruno Wolff III <[EMAIL PROTECTED]> writes:
> | The original envelope address is:
> | A B@ARPA
> | 
> | It is only encoded as:
> | <"A B"@ARPA>
> 
> Rather than arguing about what "is" means, I simply observe (again)
> that for any value of "is" qmail-smtpd wrongly transforms <"A B"@ARPA>
> into <A B@ARPA>, and wrongly behaves differently than qmail-inject.
> 
> It's not a question of encoding one thing two different, valid ways.
> It's a question of transforming a valid encoding into an invalid one.

No you are comparing encoded and unencoded values. In some cases
they are the same and some cases they aren't.

In the case you are talking about qmail isn't going from one rfc 821
encoding to a supposedly different encoding. It is just decoding the
address.

The behavior of qmail-inject is just different than sendmail. It isn't
incorrect. Sendmail expects encoded (I think our  sendmail expects rfc 822
encodings), while qmail-inject takes unencoded values. Qmail encodes the
addresses it is given, though in common cases the encoded and decoded
strings match.




I am having problems using tcpserver to log messages

I use this command to start my pop server
supervise /var/supervise/qmail/pop3d tcpserver -v 0 pop-3 /usr/sbin/in.pop3d
| setuser qmaill cyclog /var/log/qmail/pop3d/ &

This command for qmail-send
supervise /var/supervise/qmail/send /var/qmail/rc | setuser qmaill cyclog
/var/log/qmail &

and this command to start my smtp server
supervise /var/supervise/qmail/smtpd tcpserver -v -x/var/qmail/smtp/tcp.smt
        -u$QMAILDUID -g$NOFILESGID 0 smtp \
        /var/qmail/bin/qmail-smtpd-wrapper 2>&1 | setuser qmaill accustamp |
\
        setuser qmaill cyclog /var/log/qmail/smtpd &

The only one that works is the smptd server
My directories are the same permissions etc
drwxr-xr-x   4 qmaill   root         1024 Jul 19 15:32 qmail/
drwxr-xr-x   2 qmaill   root         1024 Jul 19 15:23 pop3d/
drwxr-xr-x   2 qmaill   root         1024 Jul 19 14:50 smtpd/

The extremely strange thing is all the pop messages get logged to the
console, not the pop3d directory, the qmail-send messages I never see
anywhere and smtpd logs correctly into its directory.

I used a slightly modified start from
http://web.infoave.net/~dsill/lwq.html#start-qmail

Any ideas? Please?

Tim Hunter
[EMAIL PROTECTED]
CIMx Company
1001 Ford Circle
Cincinnati, OH 45150
p 513 248-7700
f 513-248-7711
http://www.cimx.com





On Mon, Jul 19, 1999 at 03:35:10PM -0400, Tim Hunter wrote:
> I am having problems using tcpserver to log messages
> 
> I use this command to start my pop server
> supervise /var/supervise/qmail/pop3d tcpserver -v 0 pop-3 /usr/sbin/in.pop3d
> | setuser qmaill cyclog /var/log/qmail/pop3d/ &
> 
> This command for qmail-send
> supervise /var/supervise/qmail/send /var/qmail/rc | setuser qmaill cyclog
> /var/log/qmail &
> 
> and this command to start my smtp server
> supervise /var/supervise/qmail/smtpd tcpserver -v -x/var/qmail/smtp/tcp.smt
>         -u$QMAILDUID -g$NOFILESGID 0 smtp \
>         /var/qmail/bin/qmail-smtpd-wrapper 2>&1 | setuser qmaill accustamp |
> \
>         setuser qmaill cyclog /var/log/qmail/smtpd &
> 
> The only one that works is the smptd server
> My directories are the same permissions etc
> drwxr-xr-x   4 qmaill   root         1024 Jul 19 15:32 qmail/
> drwxr-xr-x   2 qmaill   root         1024 Jul 19 15:23 pop3d/
> drwxr-xr-x   2 qmaill   root         1024 Jul 19 14:50 smtpd/
> 
> The extremely strange thing is all the pop messages get logged to the
> console, not the pop3d directory, the qmail-send messages I never see
> anywhere and smtpd logs correctly into its directory.

You need a 2>&1 in there, just before the |, to redirect stderr to stdout.
Otherwise it'll write stderr (which is where the -v output goes), to the
console.

Chris




Hello all,

I'm working on a qmail-sql patch and since I want to setup a server farm, it
would make more sense to compile qmail statically. I really only want the
servers to be a bare install. Anyone know if I will see a performance hit by
compiling static? I imagine there will be more memory used.. but with the
price of ram these days.. I'm not to worried.

-j

---
Robert J. Adams [EMAIL PROTECTED] http://www.siscom.net
Looking to outsource news? http://www.newshosting.com
SISCOM Network Administration - President, SISCOM Inc.
Phone: 937-222-8150 FAX: 937-222-8153





Hello again,

We're in the process of switching to qmail but we have a cutomer who has 18
characters in their email.  qmail is only  allowing 15.  is there anyway to
change the 15 to something higher?

Is there any documentation on aliasing?



I'd like to take all of my emails from one domaing... ie [EMAIL PROTECTED],
[EMAIL PROTECTED]  etc.. and have all the emails from that domain name going
to another email address that's a different domain.

Thanks for your help.

Sienna 





On Mon, Jul 19, 1999 at 06:28:05PM -0700, Sienna wrote:

qmail doesn't really limit the size of the local part of an email
address. Perhaps the OS is limiting the length of the username, when
qmail uses the qmail-getpw call to determine if an address is local. To
test this, run:

/var/qmail/bin/qmail-getpw <long username> | tr '\0' '\n'

and see what comes back. If you do find that the username is getting
truncated, you'll have to make an entry in /var/qmail/users/assign for
this user to override the qmail-getpw call. See man qmail-users.

For more info on aliasing, read /var/qmail/doc/INSTALL.alias

> Hello again,
> 
> We're in the process of switching to qmail but we have a cutomer who has 18
> characters in their email.  qmail is only  allowing 15.  is there anyway to
> change the 15 to something higher?
> 
> Is there any documentation on aliasing?

-- 
See complete headers for more info




Hi all.

According to RFC822:

>4.3.1.  RETURN-PATH
>
>        This field  is  added  by  the  final  transport  system  that
>        delivers  the message to its recipient.  The field is intended
>        to contain definitive information about the address and  route
>        back to the message's originator.

Return-Path is added by the *final* transport system. So why is it added by
qmail-inject?

On the same subject, if I send an email message to [EMAIL PROTECTED] without
the Return-Path field (e.g. by using qmail-queue) but using VERP (so that
the envelop sender becomes [EMAIL PROTECTED]), the
email.com server writes the Return-Path as <[EMAIL PROTECTED]>
stripping everything left of the equal sign.

As far as I can tell this only happens with email.com (and other domain
names they host). Is this a problem with their mail server or should I
always include the Return-Path field, against RFC822?

Thanks.

David.
______________________________________
David Villeger
(212) 972 2030 x34

http://www.CheetahMail.com
The Internet Email Publishing Solution




On Mon, Jul 19, 1999 at 07:24:41PM -0400,
  David Villeger <[EMAIL PROTECTED]> wrote:
> 
> Return-Path is added by the *final* transport system. So why is it added by
> qmail-inject?
> 

Are you sure about that? The man page indicates that it deletes return-path
headers. It does say that it will set the envelope sender address to that
found in a return path header in some cases.




David Villeger writes:

> On the same subject, if I send an email message to [EMAIL PROTECTED] without
> the Return-Path field (e.g. by using qmail-queue) but using VERP (so that
> the envelop sender becomes [EMAIL PROTECTED]), the
> email.com server writes the Return-Path as <[EMAIL PROTECTED]>
> stripping everything left of the equal sign.

People are bitching about this on other mailing lists.  email.com is
broken.


-- 
Sam





Sam writes:
 > David Villeger writes:
 > 
 > > On the same subject, if I send an email message to [EMAIL PROTECTED] without
 > > the Return-Path field (e.g. by using qmail-queue) but using VERP (so that
 > > the envelop sender becomes [EMAIL PROTECTED]), the
 > > email.com server writes the Return-Path as <[EMAIL PROTECTED]>
 > > stripping everything left of the equal sign.
 > 
 > People are bitching about this on other mailing lists.  email.com is
 > broken.

Yeah, it's only been happening for about four days.

-- 
-russ nelson <[EMAIL PROTECTED]>  http://crynwr.com/~nelson
Crynwr supports Open Source(tm) Software| PGPok | Government schools are so
521 Pleasant Valley Rd. | +1 315 268 1925 voice | bad that any rank amateur
Potsdam, NY 13676-3213  | +1 315 268 9201 FAX   | can outdo them. Homeschool!




I've read the "tips" section at qmail.org, but I don't quit understand how to
fix this problem.

I can't send mail to anybody unless I have that person's domain listed in the
rcpthosts file.  How can this be fixed?

Quinn



===
-------------------------------------------
     ~     Quinn P. Coldiron
    . .    Deal Lab, University of Nebraska
    /V\    http://deal.unl.edu/quinn/
   // \\   [EMAIL PROTECTED]
  /(   )\
   ^`~'^
_________________________________________________________
Do You Yahoo!?
Get your free @yahoo.com address at http://mail.yahoo.com





On Mon, Jul 19, 1999 at 10:09:07PM -0400, Quinn Coldiron wrote:

FAQ 5.4

> I've read the "tips" section at qmail.org, but I don't quit understand how to
> fix this problem.
> 
> I can't send mail to anybody unless I have that person's domain listed in the
> rcpthosts file.  How can this be fixed?

-- 
See complete headers for more info


Reply via email to