How to Troubleshoot and Restart the OpenVMS Mail Utility

[craigdedo]

How do you restart the OpenVMS Mail Utility on a single server within an OpenVMS cluster? This afternoon the Mail Utility stopped working on one of the members of the cluster. The OpenVMS Mail Utility works properly on the other three (3) cluster members. I sent email messages to my Internet email address from all four (4) members of the cluster. Three (3) of the test messages arrived in my Inbox but the fourth (4th) test message did not. From this test it is obvious that the Mail Utility is not working properly on one of the cluster members.

This is important because certain production operations rely on having the OpenVMS Mail Utility able to send messages to Internet addresses properly. The intended recipients need to know of any problem reports that are sent by email.

I tried to look up configuration and troubleshooting information about the Mail Utility in the official OpenVMS documentation and came up empty. I tried doing a general search on the Internet and also came up empty.

What tests should I run? What DCL commands should I use to find and correct the problem? What utilities should I use?

Other than rebooting the cluster member, what are my options for troubleshooting the problem and fixing it?

[arne_v]

The MAIL Utility as in the mail client is a stateless standalone program that can't really break.

I assume the problem is with the SMTP service responsible for handling internet email.

That service is documented in the TCP/IP documentation (not the VMS documentation).

https://docs.vmssoftware.com/vsi-tcpip- ... #SMTP_CHAP

It provide some ideas about where to look.

[cct]

Bear in mind that VMS Mail is unable to connect to a secure mail server, as I understand it. If you mail provider has moved to secure mail only, that might have broken it.

[arne_v]

It is decades since I have seriously used SMTP on VMS, but my first thought is to do:

$ @sys$startup:tcpip$smtp_shutdown
$ @sys$startup:tcpip$smtp_startup

on the node where mail is not working.

[alexwong]

In the past I used telnet to port 25 on the mail server to verify at least connectivity is ok, might be worth a try.

[sms]

Code: Select all

   I lack a true understanding of who does what in this stuff, but...

> In the past I used telnet to port 25 on the mail server to verify at
> least connectivity is ok, might be worth a try.

   That's a plausible test to see if a server is _listening_, but it
might say little about whether anyone can/will _send_ anything.

   The manual shows some diagnostic logical names.  For example, I
define TCPIP$SMTP_RECV_TRACE, but, again, that's for incoming.

   I had a problem in my cluster where too many systems were listening
(and delivering).  Each of my cluster members has its own system disk
and UAF, and I was getting incoming mail delivered all over the place.
Eventually, I realized what was happening, and stopped the SMTP service
on all but my main system.  So, no redundancy, but my non-main systems
are too ignorant to be helpful.

   For outgoing, I'd worry about the queues, especially on the problem
system:

      show queue *SMTP*

Are outgoing messages piling up in stalled queues?

   Are all your cluster members using a common TCPIP$SMTP.CONF?

   But, what do I know?

[volkerhalle]

I'd start with $ TCPIP SHOW MAIL on the cluster node, on which TCPIP mail 'does not work'

If there are jobs pending in the TCPIP$SMTP_node* queues, I'd continue with:

$ DEFINE/SYSTEM TCPIP$SMTP_LOG_LEVEL 5
$ TCPIP STOP MAIL
$ TCPIP START MAIL

$ SET ENT/AFTER=TODAY entry_no ! entry_no of a pending SMTP job

Then look at SYS$SPECIFIC:[TCPIP$SMTP]TCPIP$SMTP_LOGFILE.LOG

Once finished with troubleshooting, don't forget to deassign the TCPIP$SMTP_LOG_LEVEL and STOP/START the TCPIP mail service.

Volker.