Ioplanar battery test failure

[Nick Dokos]

We have an RS/6000, model 320, that fails a battery test that is run
by cron. I was wondering if anybody else knows anything about this.

The test is /etc/lpp/diagnostics/bin/test_batt. It is run from cron
at 4am every morning. About two weeks ago, it started failing with
the console message:

	Ioplanar battery test failed.

Diagnostics were run and a keyboard problem was detected with the
cause shown as either the keyboard (probability 70%) or the ioplanar
(probability 30%). The battery test continued to fail but later runs
of the diagnostics showed no problem with the keyboard. Field service
replaced the ioplanar with no effect on the battery test and later
changed a (the?) battery and the battery test succeeded - for 3 days...

It has now failed two days in a row.

Questions:
a) Anybody seen this?
b) Anybody know what exactly the test is testing?
c) Can it be a software bug in the test rather than
a hardware bug?
d) I am trying to use the machine to do some work, rather
than having it in pieces on my desk. The rest of the system
seems to be working OK with miscellaneous problems that
I think are setup problems (NFS and NIS). The question is
what is the chance that this will cause catastrophic failures
in the near future?

I would appreciate any hints. Email is fine (nick at osf.org or
uunet!osf.org!nick) if you think that the question is of
limited interest.

--
Nick Dokos		(nick at osf.org)
Systems Engineering	(uunet!osf.org!nick)