rfc 1122 and ICMP SOURCE HOST ISOLATED

[bra...@isi.edu]

*> From Gabriel.M...@eng.sun.com Thu Nov 10 20:40:12 1994
*> Date: Thu, 10 Nov 1994 11:11:32 -0800 (PST)
*> From: Gabriel Montenegro <Gabriel.M...@eng.sun.com>
*> Reply-To: Gabriel Montenegro <Gabriel.M...@eng.sun.com>
*> Subject: rfc 1122 and ICMP SOURCE HOST ISOLATED
*> To: Bra...@ISI.EDU
*> Cc: Steve...@eng.sun.com
*> Content-Type *> : *> text *>
*> X-Sun-Text-Type: ascii
*> Content-Length: 929
*> X-Lines: 22
*>
*> I have a question about one of the new ICMP destination unreachable codes
*> defined in rfc1122. As editor of that document, I was wondering if you could
*> give me some information, or pointers as to where to get it.
*>
*> My question is:
*>
*> Is destination unreachable code 8 (source host isolated) usable as a
*> "hard" error? Any info on what the motivation was to define this code?
*>
*> RFC1122 specifies that HOST/NET unreachable codes are "soft" errors which
*> must be taken as hints. Codes 6-12 are defined in that document, but nothing
*> is said about whether they should be considered "hard" or "soft". This omission
*> is documented in rfc1127.
*>
*> Thanks in advance for any info you can offer,
*>
*> -gabriel montenegro
*>
*> Gabriel Montenegro (Nomadic Computing Group)
*> Sun Microsystems Inc. Email: gabriel.m...@Eng.Sun.COM
*> 2550 Garcia Ave, MS UMPK14-207 Tel: (415)786-6288
*> Mountain View, CA 94043-1100 Fax: (415)786-6447
*>

Gabriel,

I don't at the moment recall exactly what code 8 was supposed to mean
(Jon Postel, did that come from your little list?), but it seems to be
administrative rather than operational. That is, it would be returned
as a statement about the topology/configuration, not as a slightly
stronger Unreachable. I would therefore expect it to be "hard".
I am CCing the former WG mailing list, to solicit other opinions.

Bob Braden