It is getting to be that time of life for our Varian Inova friends...
A 1995-vintage Inova-400 console (10.5 MHz IF, Motorola 68K CPU) is
failing to boot.
We are monitoring the bootup progress via a "tip hardwire" window running
on a Sun workstation (the 1995 original) connected to the console
serial port.
Normal messages are seen up to the stage of the bootp request to the
host workstation where it seems to hang with repeated . . . . . . . .
Monitoring the ethernet traffic for eth1 on a Linux box (with VnmrJ
3.2-Inova) using "tcpdump -v -i eth1" it appears that bootp requests are
going out, but the bootp request contains a MAC address of 0.0.0.0 instead
of 08:00:3D:24:7C:E6 that had been working all along (as configured in the
/etc/bootptab and as printed on the CPU board itself), booting from either
the original Sun or the new Linux PC.
We've tried doing obvious things like replacing the ethernet cable between
the host workstation and the console and the ethernet transceiver. Also
tried reseating the CPU board and various connectors. Console DC voltages
are normal.
This problem started after a building power failure....at first
intermittently, now fails consistently.
It seems like the console CPU board has "lost" its configured ethernet MAC
address. I vaguely recall that this parameter is normally stored in the
NVRAM on the CPU board, presumably with other important info... and could
conceivably be lost when the on-board Li battery goes bad after 18 years.
If this diagnosis is correct does anyone know workarounds short of
replacing the whole CPU board?
(e.g. replace the battery and somehow reprogram the NVRAM...maybe via the
serial port....??? or whatever)
Thanks in advance for all your sage advice.
--Mike
-----------------------------
Michael Strain
Director, CAMCOR NMR Facility
University of Oregon
Eugene, OR 97403-1253
mstrain_at_uoregon.edu
541-346-4605 office/lab
541-556-4077 mobile
http://nmr.uoregon.edu
Received on Fri Jul 19 2013 - 15:25:33 MST