sendmail stat=Deferred

[Timothy P. Jackson]

I am running into a problem with sendmail or it could be the exchange server. At certian times during the day or night we send out a bunch of automatic emails to vendors. Some of these emails get the following message....

stat=Deferred: Connection reset by smtp.xxxxx.com.

Then they sit in the mailqueue. I can manually send them by sending a sendmail -q0 command. Does any one know why this would be happening. It does not happen to the same address every time.

Any Help would be greatly appreciated!!

Thanks,
Tim

[Steven Schweda]

> stat=Deferred: Connection reset by smtp.xxxxx.com.

Who is "xxxxx.com"? You or someone else?
One domain or many?

> Any Help would be greatly appreciated!!

You first. Hiding potentially valuable
information may not be the best approach.

[Steven E. Protter · Accepted solution]

Shalom,

Steps to relay mail to an exchange server, 1 being the most crtitical.

1) Have the Exchange Server admin configure the server to accept relay from this system by IP addresss.
2) Configure sendmail with the DS directive in sendmail to forward the mail in the sendmail.cf file.

DShostname_of_exchange_srever
DSip_address_of_exhange_server
DS
[192.168.0.55]

The final example is using the numeric ip address of the exhange server which avoids DNS issues blockng access.

After you make the change:
3)
/sbin/init.d/sendmail stop
/sbin/init.d/sendmail start

Normally sudo or root privileges are necessary for this.

SEP

[Timothy P. Jackson]

Steven Schweda........

As you know xxxxxx is the name of my mail server and it doesn't matter what it is called. There fore I put all the pertinent information in the call and was hiding nothing that was needed to solve the problem. But if it helps we can call it smtp.testmail.com. There is only domain. Sendmail is working just fine except when a large batch of emails are sent at one time.

SEP...........

As I wrote above send mail is working just fine except when large batch of emails, maybe 60 - 90 emails are sent at once or in a short period time like a minute or two. So I don't thing that it is a DNS issue, but I will change the sendmail.cf file just so that we can rule it out.


Thanks,
Tim

[Patrick Wallek]

>>As you know xxxxxx is the name of my mail server

Actually, from the information given we don't know that smtp.xxxxx.com is your e-mail server. It could easily be a server somewhere on the Internet if you HP-UX server was not set up to use a smart-relay host to send mail.

When sending the e-mail, is there any way to tell at what point these messages get deferred? Is it possible that your smtp.xxxxx.com server is being overloaded by that many messages? Is there any way you can send the 5 or 10 messages at a time, pause, send another 5 or 10, etc.?

Barring that, set up a cron job to invoke 'sendmail -q' so that the queue is cleaned out regularly.

I run 'sendmail -q' every 15 minutes on my servers that don't have sendmail running. This way the mail queue will be processed for deferred outgoing messages regularly.

[Steven Schweda]

> As you know [...]

I believe that I'm a better judge of what I
know than anyone else is.

> Actually, from the information given we
> don't know [...]

I sure didn't. With my weak psychic powers,
I'm repeatedly amazed by how little I know
about other people's situations without being
told.

> [...] Is it possible that your
> smtp.xxxxx.com server is being overloaded
> by that many messages? [...]

That'd be my guess. Some better diagnostics
(at one end or the other) might reveal what's
happening.

[Patrick Wallek]

There should be logs on the smtp.xxxx.com server as well. Perusal of those for any errors could be interesting, especially if the errors occur at the same time as your 'connection reset by' messages.

[Timothy P. Jackson]

My appologies Steven and Patrick! I should have added that this was and internal exchange server with only one domain. Because these forums are read on the open internet I hesitate to put the actual name of the mail server to decrease the chance of some hacker, with nothing else to do, from trying to mess with my exchang server. Again, my appologies!

The exchange server may be getting over loaded but according to the networking group that handles the exchange server this isn't happening. They insist that the problem is on the HP side of things. I totally disagree with this because nothing has really changed on the HP server but they are always messing with the exchange server and network.

I was finally able to get the networking group to start digging a little deeper into their logs and I believe they found the problem. They haven't told me yet what they did but the mqueue is clean this morning.

If it happens again then I will just put a cron job out there to do a sendmail -q0.

Thanks for your help!

Tim

[Steven Schweda]

> If it happens again then I will just put a
> cron job out there [...]

But, if complaining to the next fellow up the
chain worked once, ...

Better to fix the problem where the problem
really is, if possible/practical.

[Steven E. Protter]

Shalom,

Once DS setup is done, and I do recommend doing it with IP address versus hostname, using a cron job to get mail off the server is the way to go.

This avoids the difficulties of having sendmail running and worrying about vulnerabilities.

SEP