Dear all,
first of all the thank you list:
Marco Luchini
Nick James
Stephen Lowe
Nikola Milutinovic
Richard Westlake
Ricard Loken
Selden E Ball Jr
Charles Knox
Paul Roetman
Original message:
I am rather stumped by this one... a little bit of history:
* the box, a PWS 500au, suffered from the famed "high quality battery
failure" and started thinking it was an NT box,
* user thought he could fix it himself and proceeded to stick an NT
boot sector on the box by trying to install NT,
* first bit of / gets corrupted,
* given it was due for an upgrade from 4.0D PK#6 to 4.0F PK#4 I
proceed with a clean reinstall once I fix it to speak SRM again (and
upgrade the firmware to 5.7 in the meantime).
* boot from CD, partition disk, select subsets, start install...
* it starts installing OSFBASE440, copies the subset, starts verifying
and fails the verify,
* tries again automagically and fails again.
If I reboot the box it fails complaining about the lack of osf_boot so
I have the vague suspicion that perhaps (only perhaps) it is unable to
write osf_boot in the "correct place".
The CD has been tested and is 100% readable with no errors being
reported both on the dead box and another PWS - the CD drive is fine
and has been swapped for a known-good to no avail.
Stuff I've already tried:
* disklabel -z
* scu format on the disk
* scu verify (just in case the disk is dead)
* dd if=/dev/rz8c of=/dev/null (/dev/rz8c is the CD-ROM) to check if
the CD is 100% readable - no errors.
Any suggestions welcome...
[Addendum to the above said:]
In addition to my previous mail, I have now tried:
* DU 4.0E - same problem, dies on OSFBASE435,
* upgrade firmware from 5.4 to 5.7 - same problem.
I have honestly run out of ideas completely... any takers?
P.S. I've been asked if the install was UFS or AdvFS: it was UFS,
so I tried AdvFS, same result :-(
[Summary, sort of]
What happened next (suggested by people on the net):
1) disklabel -wr -t advfs rz0 RZxx - Nope
2) To test that, you might try dd'ing the CD directly to the drive's
c partition. (it'll complain about block 0, but that should be all.)
Of course the result won't be particularly useful, but it ought
to be bootable. - Nope
3) ECU utility - Nothing wrong
What happened next (my stuff):
1) tried swapping the CD-ROM drive (I thought, perhaps stupidly, that
the CD didn't like one bit of the CD, the one with osf_boot). That
didn't work either,
2) SCSI low-level format the disk. No difference.
3) SCSI destructive verify. Most beautifully undamaged disk in
history.
4) Try the disk in a different box. Perfect clean install.
5) Try booting the box with disk in 4) -> Unable to read osf_boot
So, my conclusion is that there is something ill which has nothing to
do with the SCSI subsystem or the IDE subsystem - perhaps SRM didn't
like the power cut and I have a half-witted SRM console??
Any further ideas welcome - how does one wipe an SRM clean & re-flash?
Clearly the upgrade process doesn't do so because console variables
are kept.
Ciao,
Arrigo
P.S. BTW, Digital, sorry, Compaq tell me that I am dreaming it but
will not come out and test the machine for another week (I am
on a cheapo maintenance contract).
Received on Tue Aug 01 2000 - 12:29:15 NZST