Mailinglist not sending

[Spacey@]

Hiyas!

I'm running 7.5.4 on Suse 9.3. When I'm creating a new mailinglist for a domain, everything works perfect: The manual added subscribers get their welcome eMail as well as the owner. But when anybody writes something to the list adress - nothing happens. It'll be not delivered, no reply - just nothing. Looks like the eMails to the list go to nirvana. I can't even find stuff in my logs:

from mail.info->

Feb 23 19:42:23 h6726 qmail-queue: dwlib[11575]: scan: the message(drweb.tmp.265Ieh) sent by test@test.com to test@test.de should be passed without checks, because contains uncheckable addresses
Feb 23 19:42:24 h6726 qmail: 1140720144.082831 new msg 3017562
Feb 23 19:42:24 h6726 qmail: 1140720144.083233 info msg 3017562: bytes 4303 from <test@test.com> qp 11577 uid 2020
Feb 23 19:42:24 h6726 qmail: 1140720144.208590 starting delivery 12841: msg 3017562 to local 8-test@test.de
Feb 23 19:42:24 h6726 qmail: 1140720144.209063 status: local 1/10 remote 0/20
Feb 23 19:42:24 h6726 qmail: 1140720144.222073 delivery 12841: success: Failure_to_exec_mailman_wrapper._WANTED_gid_110,_GOT_gid_104.__(Reconfigure_to_take_104?)/did_0+0+1/
Feb 23 19:42:24 h6726 qmail: 1140720144.222782 status: local 0/10 remote 0/20
Feb 23 19:42:24 h6726 qmail: 1140720144.222981 end msg 3017562

This is the only thing I can find - GID stuff.. I searched other issues here in the board but people said they get strange answear eMails - which didn't help me at all. Anyone an idea? Would be really appreciated!

Regards...

 

[akxak]

Check the file owners of the .qmail-* files in /var/qmail/mailnames/<domainname>/

I have the reproduceable problem (Plesk 7.5.4, Suse 9.3) that after creating a mailing list using the Plesk control interface, these files controlling the behaviour of qmail and integrating mailman belong to root:root instead of popuseropuser.

After correcting this, the problems are gone.

I meanwhile run an hourly script to correct this in case a domain user created a new mailing list.

Bye,

Andreas

 

[Spacey@]

Hi!

Thanks, changes both to popuser.popuser - but still doesn't work. mail.info:

Feb 25 16:50:27 h6726 qmail-queue: dwlib[20592]: mail: all addreses are uncheckable - need to skip scanning (by deny mode)
Feb 25 16:50:27 h6726 qmail-queue: dwlib[20592]: scan: the message(drweb.tmp.VsMv1Z) sent by test@test.com to test@test.de should be passed without checks, because contains uncheckable addresses
Feb 25 16:50:27 h6726 qmail: 1140882627.558244 new msg 3017283
Feb 25 16:50:27 h6726 qmail: 1140882627.558630 info msg 3017283: bytes 4302 from <test@test.com> qp 20593 uid 2020
Feb 25 16:50:27 h6726 qmail: 1140882627.667561 starting delivery 13180: msg 3017283 to local 8-test@test.de
Feb 25 16:50:27 h6726 qmail: 1140882627.668052 status: local 1/10 remote 0/20
Feb 25 16:50:27 h6726 qmail: 1140882627.678406 delivery 13180: success: Failure_to_exec_mailman_wrapper._WANTED_gid_110,_GOT_gid_104.__(Reconfigure_to_take_104?)/did_0+0+1/
Feb 25 16:50:27 h6726 qmail: 1140882627.679104 status: local 0/10 remote 0/20
Feb 25 16:50:27 h6726 qmail: 1140882627.679594 end msg 3017283

Of course domains are blanked (tested out) - not the real domains...

"funny" ->
GID 110 = doesn't exist ?!?!?!
GID 104 = popuser

 

[wagnerch]

Try looking at this thread .

 

[Spacey@]

Thanks!
I read that stuff somewhere before but wasn't 100% sure to "play around" with the user and group files. So changing these values doesn't harm anything else for the normal popusers popping their accounts? Just some bad thoughts about the normal mail service crashing down after changing that stuff

 

[wagnerch]

Did you verify in /etc/groups what the popuser gid is? What is gid 104? What is gid 110? Maybe that will help.

 

[Spacey@]

Just look 3-4 postings above where I wrote that

 

[wagnerch]

Ahh, that's probably your problem.

The gid for popuser should be:

popuser:x:110:

Of course the problem now is what is used in your /var/qmail/users/assign files?

You should be able to fix this by doing:

find / -gid 104 -print -exec chgrp 110 {} \; >/tmp/gid104to110.log

And then tweaking /etc/groups, and /var/qmail/users/assign... but I have never done it and can't guarantee anything. It could make it even worse.


BTW, Did this box have another distribution of qmail installed on it before Plesk was installed?

 

[Spacey@]

Hi!

The assign file reads:

...popuser:110:104:/var/qmail/mailnames/...

All the way 110:104

And no, it's a Strato Rootie - came up with complete installed 9.3 and the Plesk. I didn't install anything manually...

 

[wagnerch]

That is the problem. You can *try* what I suggested, or you can try to find the source for /var/qmail/bin/mm_wrapper and rebuild it with GID=104. Problem is any future upgrades of psa-qmail will likely break it again.

 

[Spacey@]

I'm still on this - I tried this so far:

1) Looking in /etc/passwd & /etc/group for the gid of popuser and changed it to 110

2) Did this: /usr/local/psa/tmp/hotfix-75051014.16/sbin/mchk --with-spam

which gave me no error. Using the other one:

/usr/local/psa/tmp/hotfix-75050926.17/mchk

gave me a segmentation fault.

3) created a file mm_wrapper.c with

#include <errno.h>
#include <sys/types.h>
#include <unistd.h>
#include <stdio.h>
int main(int argc, char** argv, char** env) {
if (setregid(110, 110) != 0) {
printf("Set right UID/GID for popuser in /etc/passd,/etc/group\n ");
return -1;
}
(void) execve("/usr/local/psa/qmail/bin/mm_wrapper-real", argv, env);
/* Should not get here */
}

4) move the old file: mv /var/qmail/bin/mm_wrapper /var/qmail/bin/mm_wrapper-real

5) compiled the new one: gcc mm_wrapper.c -o /var/qmail/bin/mm_wrapper

6) Got an error: mm_wrapper.c:12:2: warning: no newline at end of file

... but the new file does exist. Changed it to root.popuser

7) Stoped PSA, xinet, qmail and so on - restarted it.

Luckily I still can pop my server as normal. I can send to my existing mailinglist without getting an error - but the email to the list is still not delivered. But I don't get any errors anymore in my mail.info log:

Feb 28 11:56:12 h6726 qmail-queue: dwlib[9048]: mail: all addreses are uncheckable - need to skip scanning (by deny mode)
Feb 28 11:56:12 h6726 qmail-queue: dwlib[9048]: scan: the message(drweb.tmp.IJOu6v) sent by test@test.com to test@test.de should be passed without checks, because contains uncheckable addresses
Feb 28 11:56:12 h6726 qmail: 1141124172.870662 new msg 3016872
Feb 28 11:56:12 h6726 qmail: 1141124172.871083 info msg 3016872: bytes 4310 from <test@test.com> qp 9049 uid 2020
Feb 28 11:56:12 h6726 qmail: 1141124172.979911 starting delivery 2: msg 3016872 to local 8-test@test.de
Feb 28 11:56:12 h6726 qmail: 1141124172.980400 status: local 1/10 remote 0/20
Feb 28 11:56:12 h6726 qmail: 1141124172.993569 delivery 2: success: did_0+0+1/
Feb 28 11:56:12 h6726 qmail: 1141124172.994266 status: local 0/10 remote 0/20
Feb 28 11:56:12 h6726 qmail: 1141124172.994460 end msg 3016872


????

 

[wagnerch]

You shouldn't need to rebuild mm_wrapper, if you changed the gid. The problem is mm_wrapper is hardcoded with gid of 110.

Here is Plesk's Knowledgebase Answer. I think you did most of this.

 

[Spacey@]

Interesting.... yesterday - after doing my stuff & posting - no error occured. Later the evening I did the mchk tool again without spam-settings. Today I got the following error while sending to the list:

Mar 1 10:16:02 h6726 qmail: 1141204562.726084 delivery 114: success: Failure_to_exec_script._WANTED_gid_67,_GOT_gid_12./Failed_to_start_/usr/lib/mailman/mail/mailman./did_0+0+1/

Before and after exchanging my new compiled wrapper with the org. one?!

What the heck?!

 

[federal_reserve]

Sweet

Originally posted by wagnerch
You shouldn't need to rebuild mm_wrapper, if you changed the gid. The problem is mm_wrapper is hardcoded with gid of 110.

Here is Plesk's Knowledgebase Answer. I think you did most of this.

That link fixed me up. Had the same problems.

Plesk should fix this. Definetly a bug.

Running RHEL 4.

 

[xChrisx]

Mailinglists do not work

Hello I have a problem with the Mailinglists, I these provided and email addresses added, if however someone does not write these the address of the ruse passed on to the other addresses, which I can do thereby this again run?
I use the Plesk 7.5.4
greeting
xChrisx

--translated with Goggle.de--

 

[mrbister]

I'll also place a vote for a bug-fix for this issue. Anyone who has heard any more regarding this issue?

 

[Spacey@]

This helped me finally: http://forum.swsoft.com/showthread.php?s=&threadid=28909

 

[bobathtmldotcom]

Replacement for original mm_wrapper.c file.

This fixed my mailman problems on Plesk:

/* This should be the uid that owns the /var/spool/mailman/in directory */
#define MAIL_UID 110
/* This should be the GID that owns the /var/spool/mailman/in directory. */
#define MAIL_GID 41
/* After editing the above lines, you may compile this program with the
   following command-line:

   gcc -Wall -O -s -o mm_wrapper mm_wrapper.c

   Then change the ownership and mode of the resultant mm_wrapper file to
   the same as tthe /var/spool/mailman/in directory.  On my system, that
   would be accomplished by the following commands:

   chown 110.41 mm_wrapper
   chmod g+s mm_wrapper
*/
/* No user-serviceable parts below this point. */
#include <unistd.h> /* for seteuid, setegid */
#include <stdio.h> /* for printf */
int main(int argc, char** argv, char** env) {
        if (seteuid(MAIL_UID) != 0) {
                printf("Unable to set Effictive UID to %d\n",MAIL_UID);
                return -1;
        }
        if (setegid(MAIL_GID) != 0) {
                printf("Unable to set Effective GID to %d\n",MAIL_GID);
                return -1;
        }
        (void) execve(argv[1], argv+1, env);
        return 0;
}

 

[DCNet_James]

See this http://forum.swsoft.com/showthread.php?s=&threadid=27474

It was an issue months and months ago, but I got a fix from SW-Soft. I believe somewhere in the thread someone covered Linux distros. The main thing is the 7.5.4 install breaks the mailman process. The mailman wrapper in qmail is expecting a group ID of 110 for popuser. Its all very detailed and explained in that thread. Essentially I provided a small c program to be compiled and replace the existing mailman wrapper, plus instructions to fix the group/user id error.

Thx
James