kirilka@macrules.hack.ru (kirilka) writes:
> ---Dec 14 03:07:28 magus tftpd[1543]: tftpd: write: Host is down
> ---Dec 14 03:07:28 magus tftpd[1544]: tftpd: write: Host is down
> and so on, till Sparc doesn't give up.

Funny you should mention this.  I just saw this behavior yesterday
when trying to resurrect an old 16Meg SPARC-SLC as a hack.

The problem appears to be that under some conditions during network
booting that the sparc client will not answer arp requests for its MAC
address.  The tftp requests were being sent from the client but the
server was stuck in an infinite arp-request loop looking for the MAC
address of the client.  The client never answered any of the
arp-requests.  It was as if the SLC stopped answering arp-requests
after the tftp started.

I did manage to get the SLC booted by powering both the SLC (client)
and the ss5 (server) down and starting from scratch.  Power cycling
just the client didn't do the trick.

(The next problem I noticed is that the SLC was dog-slow.  My how
quickly we get spoiled. ;-) Opening an xterm was a 10 second affair.
Starting emacs caused everything to come to a grinding halt.)

-wolfgang
-- 
       Wolfgang Rupprecht <wolfgang+gnus@dailyplanet.wsrcc.com>
		    http://www.wsrcc.com/wolfgang/