Some more details on this.  I've compared the dhcpd debug info from a
working NetBSD/i386 1.3L system and the debug info from my NetBSD/sparc
1.4.1 system.

NetBSD/i386 1.3L, dhcpd server output

  DHCPDISCOVER from 08:00:2b:1e:59:c3 via ne1
  DHCPOFFER on 192.168.102.35 to 08:00:2b:1e:59:c3 via ne1
  DHCPREQUEST for 192.168.102.35 from 08:00:2b:1e:59:c3 via ne1
  DHCPACK on 192.168.102.35 to 08:00:2b:1e:59:c3 via ne1
  DHCPDISCOVER from 08:00:2b:1e:59:c3 via ne1
  DHCPOFFER on 192.168.102.35 to 08:00:2b:1e:59:c3 via ne1
  DHCPREQUEST for 192.168.102.35 from 08:00:2b:1e:59:c3 via ne1
  DHCPACK on 192.168.102.35 to 08:00:2b:1e:59:c3 via ne1

NetBSD/sparc 1.4.1, dhcpd server output

  DHCPDISCOVER from 08:00:2b:1e:59:c3 via le0
  DHCPOFFER on 192.168.102.35 to 08:00:2b:1e:59:c3 via le0
  DHCPDISCOVER from 08:00:2b:1e:59:c3 via le0
  DHCPOFFER on 192.168.102.35 to 08:00:2b:1e:59:c3 via le0

I guess I'll start poking around with tcpdump.  If the new dhcpd has
become picky about options and pathname lengths, then I'd say those are
some rather obnoxious bugs.

-brian.
--- Brian Chase | bdc@world.std.com | http://world.std.com/~bdc/ -----
   "...there were 4 vomiting cats and a hamster that was in a coma."
      -- Read on some forgotten web page while browsing one day.