Re: I crashed the SMTP server on V3r7

[PaulMmn <PaulMmn@xxxxxxxxxxxxx>]

>Does anyone has the similar experience? I manage to crash the SMTP
>server on the v3r7 and cause it to stop functioning such that no one is
>able to send any mails via SMTP for that matter.
>
>Please advise if there is a patch to fix this loop hole?
>
>--
>Ferdinand Tang


I wouldn't have called it a crash, but we've had an ongoing intermittent
problem with SMTP Mail.  We get a 'clog' in the outgoing mail stream.
Incoming mail works fine, but outgoing mail won't.

Symptoms include complaints that mail won't send.  If you look in subsystem
QSYSWRK, check the joblog for job QSMTPSRVR.  The first few entries include
the number of jobs in the outgoing and incoming queues.  If the numbers
aren't zero, you may have a clog, or the system may be slow.  Only way to
tell is to ENDTCPSVR *SMTP, then re-start it.  If the numbers go down, your
system is slow.  If they stay the same, you may have a clog.  Or the
recipient of one or more messages is temporarily unavailable; the message
may send later.

Even at those times when things seemed their worst, outgoing mail would
trickle through, sometimes after -weeks.-  IBM maintains that a single
'faulty' message should have no effect on other messages.

We've found no cure.  The only work around is to ENDMSF and ENTCPSVR *SMTP.
Set  the data area for SMTP to start and throw away the first mail item it
finds (Yes, it's lost).  Start MSF and SMTP and examine the joblog.  Repeat
until the outgoing queue is empty.  Now restart both MSF and SMTP with the
cold-start options (to flush things out totally).

After a cold start, SMTP behaves itself for a while (weeks, months).  Then
a clog may happen again.

IBM has taken dumps, logs, looked at the flight records, and poked around,
but without any definite answers.  Going back a ways, one problem was that
AOL had so many name servers, but IBM only checked the first 2.  If the
name servers you were after on AOL were down, the mail wouldn't send.  This
has been fixed.

Looking at the flight recorders is just about the only way to decipher the
problem; we just don't have the time to constantly monitor the files.

We're currently running fine (knock on wood).  However, we keep looking at
the joblog to make sure we're not clogged up again.

--Paul E Musselman
PaulMmn@ix.netcom.com


+---
| This is the Midrange System Mailing List!
| To submit a new message, send your mail to MIDRANGE-L@midrange.com.
| To unsubscribe from this list send email to MIDRANGE-L-UNSUB@midrange.com.
| Questions should be directed to the list owner/operator: david@midrange.com
+---