>>1) Do you have sendmail=YES in /etc/rc.conf?
>
> I only need that if I want sendmail to be available from the outside
> right? I don't want sendmail to be "on" and listenting for SMTP
> connections. I only want it working enough to send the emails the CGI
> generates on the same box.
>
>>2) Is there some error/backup processing in the CGI that will do the
>> right thing if sendmail is down?
>
> No idea. There isn't much in the way of docs for it.
>
>>3) Have you considered using postfix (postfix=YES in /etc/rc.conf)?
>> It's marginally faster and has generally the same external
>>interfaces.  It would probably "just work" if you turned it on
>>instead.
>
> Nope... but I'm considering it now. Although I don't know how to alter
> the cgi to work with postfix (its a compiled C program, and like I said,
>  has almost no docs, so I'd have to root thru the source code to see how
>  it decides what mail app to use... and I'm not very good with C yet)
>

My servers do something like this.  Most of the time, this works.  If, for
some reason, the mail message wedges on send (no route, whatever...) and
you don't have a sendmail process running scan the queue, the message will
just sit there.

I've solved the problem two different ways.  Both have their advantages.

1)  Set "sendmail=YES" in rc.conf, but disallow port 25 from anything but
your local hosts.  This way, you have the stress of running sendmail
without most of the facility ;-)

2)  Set "sendmail=NO" in rc.conf, and then, in crontab, put in a 'sendmail
-q' that runs at the top of every hour.  THis way, anything sitting
waiting to go out will run.

-- 
Dave Burgess
CTO and Chief Engineer, Nebraska On-Ramp
Bellevue, NE 68123