topic Re: SMTP mail fail in Operating System - OpenVMS

SMTP mail fail

Michael LaRoche — Wed, 03 May 2006 09:47:44 GMT

Hi,

SMTP mail has been working on the current machine (ds10 running Alpha 7.3-2) for awhile, suddenly just this last weekend it's not allowing mail. I checked the smtp queues and see that the server queue TCPIP$SMTP_MED_01 stopped. I tried doing a tcpip stop mail and start mail, and get %%%%%%%%%%% OPCOM 3-MAY-2006 10:41:30.28 %%%%%%%%%%%
Message from user SYSTEM on MEDDV1
%SYSTEM-F-ACCVIO, access violation, reason mask=00, virtual address=000000000A87
D415, PC=000000000017AB70, PS=0000001B

%%%%%%%%%%% OPCOM 3-MAY-2006 10:41:32.93 %%%%%%%%%%%
Message from user SYSTEM on MEDDV1
%QMAN-E-SYMDEL, unexpected symbiont process termination

%%%%%%%%%%% OPCOM 3-MAY-2006 10:41:32.93 %%%%%%%%%%%
Message from user SYSTEM on MEDDV1
-SYSTEM-F-ACCVIO, access violation, reason mask=00, virtual address=000000000000
0000, PC=00000000000714C8, PS=0000001B

%TCPIP-E-STARTERROR, error starting SMTP service
-TCPIP-E-QUEUE, error processing queue TCPIP$SMTP_MEDDV1_01

I checked the service smtp and it is enabled and started. I've tried disabling the service and even used tcpip$config to disable and stop the service for client smtp.

Any ideas?

Re: SMTP mail fail

Wim Van den Wyngaert — Wed, 03 May 2006 10:05:15 GMT

Check all log files in sys$specific:[tcpip$smtp].

Wim

Re: SMTP mail fail

Wim Van den Wyngaert — Wed, 03 May 2006 10:12:04 GMT

And read this.

With TCP/IP Services Version 5.x, corruption might occur in SMTP control files.

Intermittently, the SMTP server signals an error on a particular control file. The most common error message returned is SYSTEM-F-BADPARAM , although the problem can sometimes result in the access violation message SYSTEM-F-ACCVIO . During most of the occurrences, the symbiont process crashes, leaving a dump file. If you do not delete the queue entry for the control file that caused the crash, the entry will be the first entry processed when the queue is started after the crash, which causes the queue to crash again. If the job is put on hold and the TCP/IP management command SHOW MAIL/FULL is issued, the SHOW MAIL command either signals an error or hangs.

This problem is due to a periodic corruption that occurs in SMTP control files and happens more frequently with high-use systems (1000 or more control files processed per day) than with systems processing less mail.

To work around the problem, run the SMTP queue watcher command file TCPIP$EXAMPLES:TCPIP$RESTART_SMTPQ.COM. The header comments of this file describe its use and operation.

The following describes additional problems and workarounds:

SMTP does not recognize mail to the cluster alias (IP address or DNS name) as local. V4.x versions of the product do.
Workaround:
Add the name of the cluster alias to the local aliases file.
The ANALYZE MAIL command does not find control files stored in TCPIP$SMTP_COMMON when TCPIP$SMTP_COMMON is not pointing to SYS$SPECIFIC:[TCPIP$SMTP].
Workaround:
Manually submit the control file to the SMTP queue.

Wim

Re: SMTP mail fail

Hakan Zanderau ( Anders — Wed, 03 May 2006 10:14:12 GMT

Try to stop smtp, remove the queues and start smtp again. Start smtp will recreate the queues.

regards,

Hakan Andersson

Re: SMTP mail fail

Michael LaRoche — Wed, 03 May 2006 10:14:19 GMT

Wim - here's what the log shows when the problem started on 29 Apr:

%%%%%%%%%%%% 24-APR-2006 21:38:07.90 %%%%%%%%%%%%
%TCPIP-I-SMTP_LOGSUC, using log file SYS$SPECIFIC:[TCPIP$SMTP]TCPIP$SMTP_LOGFILE
.LOG

%%%%%%%%%%%% 24-APR-2006 21:38:08.03 %%%%%%%%%%%%
%TCPIP-I-SMTP_SYMBRUN, symbiont is running the queue TCPIP$SMTP_MEDDV1_01

%%%%%%%%%%%% 29-APR-2006 23:59:01.18 %%%%%%%%%%%%
%SYSTEM-F-ACCVIO, access violation, reason mask=00, virtual address=000000000A8B
D415, PC=000000000017AB70, PS=0000001B
WARNING: Fatal error signalled when processing CF= _DSA0:[SYS0.SYSMGR]0604292359
0092_SYSTEM-484.TCPIP_MEDDV1;1
Leaving as orphan

%%%%%%%%%%%% 30-APR-2006 23:59:01.15 %%%%%%%%%%%%
%SYSTEM-F-ACCVIO, access violation, reason mask=00, virtual address=000000000A8B
D415, PC=000000000017AB70, PS=0000001B
WARNING: Fatal error signalled when processing CF= _DSA0:[SYS0.SYSMGR]0604302359
0089_SYSTEM-4A1.TCPIP_MEDDV1;1
Leaving as orphan

%%%%%%%%%%%% 1-MAY-2006 00:00:56.82 %%%%%%%%%%%%
%SYSTEM-F-ACCVIO, access violation, reason mask=00, virtual address=000000000AA9
02F6, PC=000000000017AB70, PS=0000001B
WARNING: Fatal error signalled when processing CF= _DSA0:[SYS0.SYSMGR]0605010000
5657_SYSTEM-246.TCPIP_MEDDV1;1
Leaving as orphan

Re: SMTP mail fail

Steven Schweda — Wed, 03 May 2006 10:15:03 GMT

TCPIP SHOW VERSION

Latest ECO?

I don't know what's happening here, but a long
time ago, some badly structured junk e-mail
could explode the receiver. Some ECO fixed
that one.

Re: SMTP mail fail

Duncan Morris — Wed, 03 May 2006 10:19:25 GMT

are there any problems showing with

TCPIP ANAL MAIL

?

Re: SMTP mail fail

Wim Van den Wyngaert — Wed, 03 May 2006 10:24:42 GMT

If you don't have the restart command procedure you can put the first entry in hold and then start smtp. Could work too.

Wim

Re: SMTP mail fail

Michael LaRoche — Wed, 03 May 2006 10:43:28 GMT

I deleted the entries in the queue and deleted the associated files. I ran again the TCPIP$EXAMPLES:TCPIP$RESTART_SMTPQ.COM and the results are:

SUB_Q = "SYS$BATCH"
SUB_TIME = "+00:10:00.00"
PAGFILCNT_MIN = 1000 Hex = 000003E8 Octal = 00000001750
TERMINATE_SRCHSTR = ""
SMTP_EXEC_Q = "TCPIP$SMTP_MEDDV1_01"
SMTP_EXEC_Q = "TCPIP$SMTP_MEDDV1_02"
There are 1 SMTP execution queues.
SMTP_EXEC_Q = "TCPIP$SMTP_MEDDV1_01"
There are 1 stopped SMTP execution queues.
All of the SMTP execution queues are stopped.
JNUM = ""
%%%%%%%%%%% OPCOM 3-MAY-2006 11:36:19.99 %%%%%%%%%%%
Message from user SYSTEM on MEDDV1
%SYSTEM-F-ACCVIO, access violation, reason mask=00, virtual address=000000000A8A
D415, PC=000000000017AB70, PS=0000001B

%%%%%%%%%%% OPCOM 3-MAY-2006 11:36:22.68 %%%%%%%%%%%
Message from user SYSTEM on MEDDV1
%QMAN-E-SYMDEL, unexpected symbiont process termination

%%%%%%%%%%% OPCOM 3-MAY-2006 11:36:22.68 %%%%%%%%%%%
Message from user SYSTEM on MEDDV1
-SYSTEM-F-ACCVIO, access violation, reason mask=00, virtual address=000000000000
0000, PC=00000000000714C8, PS=0000001B

%TCPIP-E-STARTERROR, error starting SMTP service
-TCPIP-E-QUEUE, error processing queue TCPIP$SMTP_MEDDV1_01
%%%%%%%%%%% OPCOM 3-MAY-2006 11:36:22.71 %%%%%%%%%%%
Message from user J004042 on MEDDV1
_MEDDV1$TNA65:, TCPIP$RESTART_SMTPQ Error restarting SMTP queues: %NONAME-E-NOMS
G, Message number 176488FA

Re: SMTP mail fail

Michael LaRoche — Wed, 03 May 2006 10:52:17 GMT

Current version of tcpip is V5.4 - ECO 4.

I've done the anla mail - no errors and as stated I've stopped and restarted the smtp service and mail under tcpip.

Re: SMTP mail fail

Hakan Zanderau ( Anders — Wed, 03 May 2006 11:20:02 GMT

Try

$ DEFINE/SYS/EXEC TCPIP$SMTP_LOG_LEVEL 5
$ TCPIP STOP MAIL
$ TCPIP START MAIL

Then check the logfile
SYS$SPECIFIC:TCPIP$SMTP_LOGFILE.LOG

regards,

Hakan Andersson

ps Don't forget to reset (remove) the logical and restart smtp again. It will make SMTP to write A LOT to the logfile and will REALLY slow things down. ds

Re: SMTP mail fail

Michael LaRoche — Wed, 03 May 2006 11:24:05 GMT

The one thing I do see is that the TCPIP$SMTP_SYMBIONT.EXE is not there and when it's attempted to restart it it disappears right away, so that's why the access-vio.

Re: SMTP mail fail

Hakan Zanderau ( Anders — Wed, 03 May 2006 11:42:02 GMT

What happens if you run the symbiont manually ?

$ mc tcpip$smtp_symbiont

Do you get more than....

$ mc tcpip$smtp_symbiont
%SYSTEM-F-IVLOGNAM, invalid logical name
$

Hakan

ps. It was easier working inhouse (Digital/Compaq/HP), having access to all databases..... ;-) ds

Re: SMTP mail fail

Volker Halle — Wed, 03 May 2006 11:42:28 GMT

Michael,

the current version is TCPIP V5.4 ECO 5 - but there is nothing in the release notes regarding problems like this, though you never know...

The TCPIP$SMTP_SYMBIONT seems to be crashing with access violations when processing mails. Are there any process dumps in SYS$MANAGER:TCPIP$*.DMP ?

Consider to read the chapter about SMTP logging and enable it to try to find out, which operation is causing the ACCVIOs:

http://h71000.www7.hp.com/doc/732final/6526/6526pro_039.html#logicals_sec

Volker.

Re: SMTP mail fail

Volker Halle — Wed, 03 May 2006 11:48:09 GMT

Michael,

The one thing I do see is that the TCPIP$SMTP_SYMBIONT.EXE is not there and when it's attempted to restart it it disappears right away, so that's why the access-vio.

It's the other way around ;-)

When the symbiont starts executing (when starting the queue), it gets an ACCVIO due to some software problem and therefore exits. Check for dumps...

Volker.

Re: SMTP mail fail

Michael LaRoche — Mon, 15 May 2006 09:24:14 GMT

The problem is dealing with relay agents on a different site that the company uses. All the answers and help are greatly appreciated. I plan on putting the most current patch in any way just to get up to current run levels. Thanks to all who responded.