>Can you tcpdump(8) the outgoing connection to check whether sendmail is in
>fact timing out on a DNS query and what it is actually trying to resolve?

Not long after I sent the other message, I noticed the -v verbose option 
in the sendmail man pages (something I must have gone thru a dozen times 
looking for things to try... I just couldn't absorb it all at once).

When I tried it in verbose mode, I noticed that it sat for 60 seconds, 
then came back and complained that my hostname wasn't valid, then 
continued and sent the email (showing the communications with the mail 
server).

I had seen notes about this being a potential problem, so I looked up how 
to disable sendmail's DNS checking, and I thought I had done it. I also 
had sendmail set to just queue the messages for later delivery, and 
figured that should also prevent the delay from happening.

I guess I was wrong. As soon as I changed my hostname in my rc.conf to 
something that has a valid DNS entry, sendmail started processing and 
sending the email as soon as the <crlf>.<crlf> was hit.

I'll go back later and change my hostname to something I want it to be, 
once I put thru a new DNS entry for it.

The odd thing is, I have another mac68k NetBSD machine, and it doesn't 
seem to have this problem. It is 1.5.2 with whatever version of sendmail 
that comes with that version of NetBSD. It too has a hostname that isn't 
valid, but it doesn't have the delay in sendmail. I also noticed that it 
doesn't attempt to use just the hostname, but rather prepends the 
hostname with the machine's domain. That might make it a valid DNS entry 
as that domain might have a wildcard entry. For some reason the trouble 
machine doesn't do that hostname.domain construction and tries to use 
just the hostname.

For now, I am chalking it up to a difference in the versions of sendmail. 
Since I have it working, figuring out exactly why it doesn't act the same 
way as another machine can be left for a day that I have nothing to do.

-chris
<http://www.mythtech.net>