In message <20011217132013.A21657@WintelKiller.T> I write:

|> sd1(esp1:1:0): esp1: timed out [ecp 0xf03d68d0 (flags 0x3, dleft bc00, stat 0)], <state 1, nexus 0x0, phase(l 0, c 100, p 7), resid bc00, msg(q 0, o 0) >

Set up a spare Sparc-2 to run NetBSD-1.5.1 diskless, connected DNES to
on-board controller, and got:

sd0(esp:0:1:0): esp0: timed out [ecp 0xf03ac0d0 (flags 0x3, dleft d200, stat 0)], <state 1, nexus 0x0, phase(l 0, c 100, p 7), resid d200, msg(q 0, o 0) >

|> Next thing is to reconnect the DNES to the OpenBSD box.

Works. I can newfs, then dump | restore some stuff, fsck, it works.
This is using the same cable, terminator, and power outlet where it
didn't work five minutes before connected to the NetBSD box.

|> (could it be that the disklabels are compatible enough that NetBSD
|>  recognizes the layout but incompatible enough to cause such problems?)

To rule out this to some degree, I re-labelled the disk on SunOS, then
disklabel'ed and newfs'ed it again on NetBSD. No changes except for the
dleft/resid value.

The last straw to cling to is to get the above mentioned Sparc-2 to
netboot OpenBSD. The the hardware will not only be equal but identical
and the only difference will really be the OS...

	Volker