qmail - Sorry,_I_wasn't_able_to_establish_an_SMTP_connection._(#4.4.1)

[JC_SON]

I have been having serious problems with qmail on Plesk 7.5.4 (again)

Problem:

Sorry,_I_wasn't_able_to_establish_an_SMTP_connection._(#4.4.1)

I have :

1. Removed DrWeb
2. Re installed qmail
3. Tested a telnet connection from inside and outside - both refused
4. Checked that qmHandle can read the queue's
5. Checked smpt.psa in xinetd.d
6. Check that qmail is running - ps aux..
7. /etc/init.d/qmail status - reports status as stopped !!!!!
8. Munched a banana
9. Smelled my sneakers

Anybody have any ideas!
Thanks

 

[sieb@]

What do the logs say?

 

[JC_SON]

Sieb

Logs with a restart:

Sep 22 10:20:58 vps537 qmail: 1127377258.231024 delivery 3: deferral: Sorry,_I_wasn't_able_to_establish_an_SMTP_connection._(#4.4.1)/
Sep 22 10:20:58 vps537 qmail: 1127377258.231066 status: local 0/10 remote 1/20
Sep 22 10:20:58 vps537 qmail: 1127377258.240135 delivery 1: deferral: Sorry,_I_wasn't_able_to_establish_an_SMTP_connection._(#4.4.1)/
Sep 22 10:20:58 vps537 qmail: 1127377258.240181 status: local 0/10 remote 0/20
Sep 22 10:23:25 vps537 qmail: 1127377405.165559 status: exiting
Sep 22 10:23:25 vps537 spamd[1284]: server killed by SIGTERM, shutting down
Sep 22 10:23:25 vps537 spamd[1271]: server killed by SIGTERM, shutting down
Sep 22 10:23:25 vps537 spamd[1282]: server killed by SIGTERM, shutting down
Sep 22 10:27:17 vps537 qmail: 1127377637.653367 status: local 0/10 remote 0/20
Sep 22 10:27:20 vps537 spamd[3353]: server started on port 783/tcp (running version 2.63)
Sep 22 10:27:21 vps537 spamd[3365]: server started on UNIX domain socket /tmp/spamd_full.sock (running version 2.63)
Sep 22 10:27:22 vps537 spamd[3367]: server started on UNIX domain socket /tmp/spamd_light.sock (running version 2.63)
Sep 22 10:28:30 vps537 qmail: 1127377710.650156 starting delivery 1: msg 12861813 to remote [email protected]
Sep 22 10:28:30 vps537 qmail: 1127377710.650214 status: local 0/10 remote 1/20
Sep 22 10:28:30 vps537 qmail: 1127377710.650384 starting delivery 2: msg 12861815 to remote [email protected]
Sep 22 10:28:30 vps537 qmail: 1127377710.650411 status: local 0/10 remote 2/20
Sep 22 10:28:30 vps537 qmail: 1127377710.659449 delivery 2: deferral: Sorry,_I_wasn't_able_to_establish_an_SMTP_connection._(#4.4.1)/
Sep 22 10:28:30 vps537 qmail: 1127377710.659502 status: local 0/10 remote 1/20
Sep 22 10:28:30 vps537 qmail: 1127377710.659640 delivery 1: deferral: Sorry,_I_wasn't_able_to_establish_an_SMTP_connection._(#4.4.1)/
Sep 22 10:28:30 vps537 qmail: 1127377710.659677 status: local 0/10 remote 0/20
Sep 22 10:31:30 vps537 qmail: 1127377890.650136 starting delivery 3: msg 12861814 to remote [email protected]
Sep 22 10:31:30 vps537 qmail: 1127377890.650197 status: local 0/10 remote 1/20
Sep 22 10:31:30 vps537 qmail: 1127377890.650371 starting delivery 4: msg 12861817 to remote [email protected]
Sep 22 10:31:30 vps537 qmail: 1127377890.650400 status: local 0/10 remote 2/20
Sep 22 10:31:30 vps537 qmail: 1127377890.661379 delivery 3: deferral: Sorry,_I_wasn't_able_to_establish_an_SMTP_connection._(#4.4.1)/
Sep 22 10:31:30 vps537 qmail: 1127377890.661434 status: local 0/10 remote 1/20
Sep 22 10:31:30 vps537 qmail: 1127377890.662094 delivery 4: deferral: Sorry,_I_wasn't_able_to_establish_an_SMTP_connection._(#4.4.1)/
Sep 22 10:31:30 vps537 qmail: 1127377890.662138 status: local 0/10 remote 0/20

 

[jamesyeeoc]

Originally posted by JC_SON
I have been having serious problems with qmail on Plesk 7.5.4 (again)

Problem:

Sorry,_I_wasn't_able_to_establish_an_SMTP_connection._(#4.4.1)

I have :

1. Removed DrWeb
Fine.

2. Re installed qmail
Did you use the --force option? (rpm -Uvh --force psa-qmail......)

3. Tested a telnet connection from inside and outside - both refused
You did test trying to connect to port 25 and 110, not just regular telnet port, right?

4. Checked that qmHandle can read the queue's
That's fine, that means the directories exist.

5. Checked smpt.psa in xinetd.d
Had you edited the file previously? If you did, try restoring a backup copy just in case the editor may have messed with the long line(s).

6. Check that qmail is running - ps aux..
Which exact processes were listed as being in memory?

7. /etc/init.d/qmail status - reports status as stopped !!!!!
This probably means that qmail-send was not in the process list in #6. Normally would report "qmail-send (pid nnnn) is running..."

8. Munched a banana
I love good ripe bananas....

9. Smelled my sneakers
No comment.

Your other post was in a thread where the problem was related to a VPS server, are you also running on a VPS/Virtuozzo type server, or are you on a standalone/dedicated??

You did not mention checking /var/log/messages for any indication of qmail loading errors, or any errors for that matter.

You did not mention checking your firewall settings, or temporarily disabling the firewall to see if that may be causing a connection failure.

You did not mention if webmail was working on the server or not, just curious to see....

Did you try using netstat to see if it shows anything listening on ports 25 or 110?

netstat -an |grep ':25'
(should show: tcp  0   0.0.0.0.0:25  0.0.0.0:*  LISTEN) or similar

netstat -an |grep ':110'
(should show: tcp  0   0.0.0.0.0:110  0.0.0.0:*  LISTEN) or similar

To get more detail, replace the -an with -anvep

Hope this helps lead you to finding the cause... Cross posted again, now I'll look at your logs...

Anybody have any ideas!
Thanks

 

[JC_SON]

2. used Force
3.1 used telnet localhost 25 - connection refused
3.2 used telnet locahost 110 - connected ok
5. no
6. ps waux | grep qmail - as required for qmail to operate

qmails    8710  0.0  0.0  1452  356 pts/1    S    11:07   0:00 qmail-send
qmaill    8712  0.0  0.0  1412  412 pts/1    S    11:07   0:00 splogger qmail
root      8713  0.0  0.0  1412  312 pts/1    S    11:07   0:00 qmail-lspawn ./Maildir/
qmailr    8714  0.0  0.0  1408  340 pts/1    S    11:07   0:00 qmail-rspawn
qmailq    8715  0.0  0.0  1400  296 pts/1    S    11:07   0:00 qmail-clean
root      9397  0.0  0.0  1504  448 pts/1    S    11:16   0:00 grep qmail

Also
1. did a netstat - no SMTP ports
2. webmail can be accessed by cannot send - pear socket error reported (thats where I noticed it first)
3. checked chmod u_s state for qmail-queue as DrWeb/Plesk does not set this flag correclty
4. sacrificed a rubber chicken
5. Firewall shoud not be an issue - but will check

Further:
1. It is a VPS
2. Nothing was changed on this system

 

[jamesyeeoc]

Have you uninstalled or disabled WatchDog? If not, you should...
http://forum.sw-soft.com/showthread.php?threadid=26689&highlight=qmail+AND+VPS+AND+cron

There was also a thread which dealt with a "a1dummy-psa" file which runs as a cronjob every 10 minutes to see if qmail had anything to do, if not then it would shutdown qmail. It is buggy and should be disabled.
http://forum.plesk.com/showthread.php?threadid=25814

 

[JC_SON]

1. theres a cron job to check mail every 10 mins - qmail_check - nothing odd there

- took it out in any case

2. Watchdog is not installed
3. Found the dummy

 

[jamesyeeoc]

Read the other thread about the a1dummy, I don't remember if there was anything else to do other than disabling the cronjob.

 

[JC_SON]

Removed the cron and rpm -e the files a1dummy package

I see there is a also a 0adummy package

 

[jamesyeeoc]

I have no clue what the 0adummy package is....

 

[JC_SON]

Still at a loss for mail

Checked iptables -L - nor rules found

 

[JC_SON]

Problem solved:

Seems that xinted expected a file call smtp_psa while smtp.psa in the dir.

 

[jamesyeeoc]

Funny how the eyes play tricks on us. When I just re-read your post, in number 5 I didn't even notice you had typed 'smtp.psa' and what my eyes saw was 'smtp_psa' which is the correct filename, else I would have commented on that much earlier...

 

[JC_SON]

The 64 000 banana question is how did that happen?

 

[jamesyeeoc]

My crystal ball is offline at the moment, but it may have been when you were uninstalling/reinstalling packages, or maybe even a typo when modifying/copying/moving the smtp_psa file when you were trying to solve the problem. If you're like me, and work too many hours until you can't see straight, it becomes quite easy to make simple typos and such.

Have you found out what the 0adummy package is? Maybe related to the a1dummy?

 

[JC_SON]

No - but I decided it would be wise to leave the 0dummy alone (for now).

And on the copying of the smtp_psa file - not me - so somewhere in the upgrade/re-install process from 7.1.x to 7.4.x it happended.

 

[sadara]

Originally posted by jamesyeeoc
Funny how the eyes play tricks on us. When I just re-read your post, in number 5 I didn't even notice you had typed 'smtp.psa' and what my eyes saw was 'smtp_psa' which is the correct filename, else I would have commented on that much earlier...

thanks a lot, people. you've just solved a problem for me.

two questions:

1. do you have any idea how smtp_psa gets to be renamed to smtp.psa?
2. i'm assuming that smtps.psa should be renamed to smtps_psa, too. is this right?

again, thanks a bunch.

 

[spot]

Originally posted by sadara
1. do you have any idea how smtp_psa gets to be renamed to smtp.psa?
2. i'm assuming that smtps.psa should be renamed to smtps_psa, too. is this right?

i don't know what was the name before, but today it was smtp.psa. I enabled Watchdog last night and since then qmail would not start, even when i stopped watchdog, even when I uninstalled it !!!! Qmail refused to start.
I checked to see if these 2 files where renamed, and yes, they were .psa and not _psa.
I changed them back, reboot, works ok, and I think right now its is sending to my mailbox mail that where received those "dead" hours - I am not sure about this yet, but surely I can send and receive mail now.
Dimitris

 

[jamesyeeoc]

Yes, both files should have the "_" instead of the period "." in the filename. I believe it is a bug in one of their updates, but never spent the time to track it down totally. As a precaution, I put a script in cron to check/rename the files, just in case... it may not be needed, but didn't want to find out the hard way

 

[sadara]

looking back through my support notes, i've come to the conclusion that it was the/a plesk 7.5.4 installer which was responsible. the problem with mail began just after a problem with the plesk cp was fixed. not exactly conclusive proof, i admit, but none of the other explanations / theories i've come across explain the phenomena better.