Networking help

[pll at lanminds.com]

In a message dated: Mon, 16 Dec 2002 20:50:31 EST
bscott at ntisys.com said:

>On Mon, 16 Dec 2002, at 2:58pm, pll at lanminds.com wrote:
>> While I'm inclined to agree, the question I'm still trying to answer is,
>> why did work fine on some systems and not on others.
>
>  Paul, you and I both know that computers have enough trouble working when
>everything is configured right.  When you're doing something that's
>obviously stupid, you shouldn't be surprised if demons fly out of your
>nose[1].

Right, I agree with that.  My point was that it is not obvious that 
this is stupid, given that Linux is the only OS which allows for 
*multiple* default routes!

When I first saw this behavior my initial reaction was "Ooooh, that's 
really bad!", but somewhere along the line, I was given to believe 
that it on Linux it is not only okay and expected, but that "Linux 
does the Right Thing(tm)!".  So, I was a victim of misinformation, as 
well as not trusting my gut instinct!

(as for nasal demons, I'm intrigued by the jargon file's explanation. 
I was almost positive that it was a reference to Ren & Stimpy, but 
then I remembered upon reading the JF, that R&S referred to Nose 
Goblins[1], not 'nasal demons'.  I also assume from the spelling, that 
these are bad 'demons', and not the good 'daemons' we're used to in 
UNIX land ;)

>> Additionally, why was it sporadic, and why did it eventually work? I.e.,
>> if a system were in the state where it was not responding to pings, after
>> about 20-30 seconds, it would start answering them.
>
>  This is just a guess, but the timeout on ARP is 20 seconds, IIRC.

Interesting.  I didn't know that!

>Maybe it took that long for the computer to realize 'T' wasn't actually there, and
>try routing around the failure [2].

Possibly, that's the conclusion I'm leaning towards.  However, let's 
assume 'T' did exist.  So what?  IMO, the icmp packets should never have 
gone to 'T' anyway.  Shouldn't have gone back out the interface 
they came in on? (this seems logical to me, but then, I think I've 
pretty well established the fact that I don't understand networking 
at the protocol layer :)

>> (Btw, I refuse to admit that you were right ;)
>
>  I dunno, Paul.  I have to side with Derek on this one.

That remark was obviously tongue in cheek.  Derek and I have a long 
history of him being right and me refusing to admit it.  It was more 
an inside comment to Derek than a public refusal to believe him.

>Here I was, all along, giving you the benefit of the doubt.

There was your first mistake.  Please note the times of my initial 
posts on this topic.  All before noon, all before the required dosage 
of caffeine had entered my bloodstream :)

> I said to myself, "Paul's a seasoned system administrator. He couldn't
> possibly be dumb enough to configure a gateway that doesn't exist." 

Seasoned sysadmin, yes.  Completely knowledgable of the technical 
details of networking protocols, no.  Easily confused by the things 
Linux does differently than Solaris, yes.

Also, I find it interesting that defining a 'gateway' statement for 
an interface in Linux leads to a default route being configured 
despite one already existing.  Under Solaris, there's a
/etc/defaultroute file where you define *the* default route.
(you may be able to define more than one in this file, I don't know, 
I never tried, and it's been some time since I played with Solaris).

Defining a gateway in an interface config file, IMO, shouldn't lead 
to defining a default route.  At the most, it should lead to defining 
a network specific route such as:

	192.168.10.0   192.168.10.1  255.255.255.0 eth0

or something like that, but definitely not a default route.  As 'mod' 
pointed out, the definition of the world 'default' essentially means, 
when there are no other options, go this way.  Not, when there are no 
other options, choose one of the following :)

>I even said that in one of my posts!

Well, yeah, but anyone who knows me know that I have a faulty 
'vgrep'[2] !

>And here you are, trying to route packets over air.  :-)

No, I wasn't trying to route them over air, I wanted them to go in 
circles on the 192.168.10.0 subnet ;)

Footnotes:
----------

[1]	Ren & Stimpy had an entire episode about Stimpy's 'Nose Goblin'
	collection, which he "kept" under the kitchen table.  I 
	believe this may also have been the episode with "Powdered 
	Toast Man", but it's been something like 10 years since I 
	watched this classic show, and it was usually in a quite 
	hung-over state :)

[2]	'vgrep' is a term coined by Tom Buskey while we worked 
	together in reference to the fact that I'm quite prone
	to completely miss that which is spelled out clearly 
	right in front of me.  It seems to be chronic, as I
	can't remember ever having a correctly working 'vgrep' :)

-- 

Seeya,
Paul
--
Key fingerprint = 1660 FECC 5D21 D286 F853  E808 BB07 9239 53F1 28EE

	It may look like I'm just sitting here doing nothing,
   but I'm really actively waiting for all my problems to go away.

	 If you're not having fun, you're not doing it right!