[ILUG] Broadband querry

[Rick Moen]

Quoting Brian Foster (blf at utvinternet.ie):

> and there's other reasons:  reproducibility is one.

_Huh?_  How exactly does relying on some basically unknown third-party
script aid "reproducibility"?  I'm curious about this one.  How can
relying on some stranger's facility entirely outside one's control
improve one's ability to guarantee results?

Perhaps this is some sense of the term with which I was previously
unacquainted?  ;->

> another is when Rick's prerequisites are not met, but the tester is
> either curious or clewed-in enough to know an "open relay" in the wild
> is a bad idea.
>
> as someone (also Niall?) pointed out a few days ago, Rick seems to
> have this charming but exasperating habit of assuming everyone is not
> only an admin of a 24x7 cluster of (large?) servers, but also fully
> conversant in whatever gizmo happens to be under discussion.
> 
> both assumptions are obviously nonsense.

[I assume your reference must be to that interminable "Porting MyDoom" 
thread, which I was obliged to set aside unread because I didn't have
time for such things.  I haven't yet, either.  Sorry.]

Hmm, I do sympathise with the desire for simple tests, but you speak as
if it were somehow a major challenge to learn how to configure an MTA's
relaying functions, and how to type SMTP commands into somebody's port
25.

Guess what?  It's just not -- not even if you make the signal eror of
choosing to use sendmail.  I learned how to type SMTP commands by
spending about 20 minutes typing "help" into sendmail's SMTP parser, and
experimenting with amusingly forged messages to and from myself.  And
figuring out the relaying controls on your MTA is typically pretty
damned simple.

The people who get those things wrong are those who simply haven't
tried.  Sorry, but your false dilemma doesn't fly.

> apologies, Rick, if that sounds a bit rude, it is not meant to be!

No offence taken, Brian.

> but I am just a tiny bit exasperated by phrases like "aw, c'com ...
> you need to know the definition of ... and a handful of [ commands ]
> ..."

No:  It's simply not difficult.  Look, let's try to cover the essence of
this matter _right here_:  Relaying means being willing to handle mail
that's both not from your domain _and_ not to your domain.  You test the
MTA by crafting a mail of that sort.  If the MTA accepts and delivers
it, then it relays.  If the MTA doesn't, then it doesn't relay.

So, you telnet into the host's port 25.   You greet it with 
"HELO <your fully qualified hostname>".  You can now send it any SMTP
commands of your choice, such as:

MAIL FROM: sender at imaginarydomain.com   # You identify an ostensible sender
RCPT TO: recipient at anotherdomain.com    # You identify where you want it sent

If the MTA gives you a 250 acceptance message of some sort, then it's
purporting to relay it for you.  If "recipient at anotherdomain.com" is a
mailbox you can check (e.g., a yahoo account), then you can also see if 
the MTA actually _does_ deliver that mail as it claims it's doing.

> in any case, I found the instructions I posted useful
> for checking my own machine....

Better would be to log to disk the steps that Paul Vixie's script says
that it's taking, and carry out some version of them yourself.

> ( and like Niall, I also don't grok why that site suggests running its
> tests as the superuser:  can anyone guess why? )

On a *ix host, there's no reason I can imagine why one would have to
telnet out to a test script as the superuser, but maybe broken operating
systems from Washington State in the USA have a problem in that
direction.

> please note, I am not suggesting astrology is a science, or a subject
> for anything other than continuous ridicule.  similar to M$ or
> Windross.

Well, as a Capricorn, of course you'd feel that way.  ;->

-- 
Cheers,
Rick Moen                      Linux for Intel:  Party like it's 2037!
rick at linuxmafia.com