Path: utzoo!attcan!uunet!samsung!zaphod.mps.ohio-state.edu!ub!boulder!grunwald
From: grunwald@foobar.colorado.edu (Dirk Grunwald)
Newsgroups: comp.unix.aix
Subject: Flakey RS/6000 Ethernet
Message-ID: <24430@boulder.Colorado.EDU>
Date: 7 Aug 90 16:40:03 GMT
Sender: news@boulder.Colorado.EDU
Reply-To: grunwald@foobar.colorado.edu
Distribution: comp
Organization: University of Colorado at Boulder
Lines: 25


Previously, I posted a couple of questions w.r.t. the RS/6000. I
finally got 'amd' to compile & work with no problems. The remaning
problem is the worst, and might people frown on the machine heavily
(we're thinking of buying several with a NSF educational computing
grant & a recent CISE-II grant).

When we power-up the RS/6000, it will remain on the network for a
while.  Eventually, with no action on our part, it will not connect
with hosts connected to the other side of a retix bridge. It will
connect with hosts on its side, but not the other hosts. If you flush
the arp table entries, it will not populate them w/the ``other side''
hosts again. It effectively appears to not be able to broadcast over
the retrix.

We set the broadcast address to xx.yy.zz.255 and the netmask to
0xffffff00 and enabled arp on en0. No help. Rebooting doesn't help.
Power cycle does help, sometimes.

Has anyone seen this problem? So far, I haven't been *that* impressed
with the performance for non-floating point applications. Network
hassles don't help.

Dirk Grunwald -- Univ. of Colorado at Boulder	(grunwald@foobar.colorado.edu)
						(grunwald@boulder.colorado.edu)