inter-vendor operability of rdump...

[Ross Alexander]

rds95 at leah.Albany.Edu (Robert Seals) writes:
[essentially, he can't get an AIX box to dump to an Ultrix 2.2 muVax]
[and that the diagnostic is "lost connection to remote host"]

I, too, am experiencing some problems along these lines.  I recently
got a DS2100 dropped into my lap (smiles modestly, waves to the crowd)
and naturally I started hooking things to it.  An Exabyte EXB-8200 rev
2502 was amongst the first sucesses.  Naturally, my thoughts turned in
the "hey, this would make a great DUMPHOST" direction.

And, for a few blessed runs (under Ultrix 4.0&patch) it worked.  Like
a charm.  I naturally went out and had a beer.

The next day, I built a new kernel (in an effort to trim the fat off
the generic one) and rebooted.  Suddenly, no more dumphosting... the
remote machines (all suns 3s & 4s) kept reporting "lost connection to
remote machine - broken pipe".  Yes, I patched /etc/ttys, I set up
/.rhosts, "rsh <DS2100> <whatever>" runs fine from the Suns,
/usr/etc/rmt seems to be present, executable, & undamaged.  The really
bizzare thing is that other Ultrix systems (2.3 versions) can dump to
the DS2100 iff they use the rdump -o flag, but Sun boxes just can't
make it work at all.

Does anyone have a handle on this?  My current workaround is (command
executed on a sun host):

	sun.box# dump 0sdf 5000 150000 - /dev/XXXXX \
		 | rsh ds2100.box dd of=/dev/rmt0h obs=64b

which works just fine, but leaves much to be desired in terms of error
handling and the like.

-- 
--
Ross Alexander    rwa at cs.athabascau.ca    (403) 675 6311    ve6pdq