> So they try again - and by this time the cache has more info,
> either enough for it to succeed, or at least get a lot further (and
> then perhaps third time lucky).
(I got it on the third try, starting from nothing cached.)
But let me bring up a related problem ...
I've had occasions of users (and servers) here that couldn't
communicate because they had screwed up their default route and/or
netmask. Obviously setting these parameters is an error-prone
procedure that leads to operational instability. Even DHCP is not a
reliable solution because the server gets its information from human
hands.
What I've implemented on all our systems and routers is much more
robust and I think it should be added to a new rev of the host &
router requirements docs immediately:
Every host sending a packet SHALL encapsulate that packet using
the appropriate IP-in-IP header and send it to the all-routers
multicast address. Routers receiving such a packet SHALL
decapsulate and forward it.
Sure, this leads to some duplicatation, but that's allowed by the IP
spec. It does prevent the occasional configuration error from
stopping the communication.
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page: http://playground.sun.com/ipng
FTP archive: ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------
