Sendmail mail error when Exchange server is down

[kclute]

OS AIX 5.1 ML02
Sendmail 8.11
When sending mail to our exchange server under normal conditions, mail works fine. When the exchange server is down such as when rebooting any mail sent to the server gets the following returned error:
----- The following addresses had permanent fatal errors -----
user@company.org

----- Transcript of session follows -----
554 5.0.0 Can't connect to address family 0
550 5.1.2 user@company.org... Host unknown

And the email is not left in the mail queue to be resent during next queue run. Is there anyway of having sendmail queue these messages until the send is ok? We are a hospital and these emails are not easily reproduced RX orders, lab results and such that need to send!

Thanks
Kirk

 

[Stingreen]

Are you running DNS and exchange server together ?

 

[kclute]

If you mean, are the DNS and exchange server on the same box? The answer is no. The DNS servers are separate systems from our mail servers.
And if I ping company.org it successfully pings the name server and if I ping the mail server while the mail server is down it does return the correct ip address and times out.
Any thoughts?

Thanks
Kirk

 

[Stingreen]

Let me clarify how this works so you can get some ideas where the real problem is.

Whenever you send an email to the remote recepient, the email server checks FIRST if remote domain does have an MX record or not, not Host record. If MX exists, then there is no way your mail server could bounce back as "host not found" . It will run the queue until "queuereturn" value is satisfied. There is 1 possibility which your DNS server may not cached the domain records properly, Therefore DNS thinks that there is no record and bounces the mail back.

>if I ping the mail server while the mail server is down it >does return the correct ip address and times out.

What do you mean? ofcourse it will timeout because it is down.

Email me if you want further information or you can just port your domain and mail server address here and I'll check that out for you,
perltech@mynet.com

 

[kclute]

Thanks for the info Stingreen, but I believe I may have found the fix for our problem. Searching IBM's AIX web site I located what seems to be a very promising fix to their code.
Quote from the IBM site.
"APAR IY33426 - Sendmail fails to queue message if host not reachable"

I will be installing this fix and hopefully this will do the trick.

Thanks again
Kirk