SUMMARY: Sun Ultra 1 / partition messed

tommi.ripatti@siemens.fi
Fri, 20 Feb 1998 11:12:35 +0200

Again, hard work proved to be the only solution. No tricks helped here.

Ultra1 should definitely be reseted before using the eeprom diag
commands (probe-scsi-all...), this goes probably for all Sparcs
although SS20's don't usually freeze like this.

Thomas Anders <anders@hmi.de> had suggested using a peace
of pd software called "scsiinfo" instead of eeprom
diagnostic features.

As to fixing the filesystem, I booted from CD-ROM, fsck'd all mounts
and mounted everything.

Everything else was OK, but the root partition's file hierarchy was lost.
I think all files were in the lost+found directory, so although all
dir/file
names in the root it self were gone, inodes under that were uncorrupted.

I worked through the /etc and salvaged all configuration data.

Then I just backuped everything else with ufsdump and reinstalled OS
from scratch.

Nothing fancy, just work. Could've probably gone through this with less
work, but I chose to do it in a 100% safe way.

Thanks for all ...

Thomas Anders <anders@hmi.de>
Jim Musso <jam@cdicad.com>
Loring Safford <lsafford@mitretek.org>
Ian Wallace <iwallace@bcoe.bm>
Dave... <wanamaker@Radix.Net>
Eddy Sutanto <sutantoe@jakarta.geoquest.slb.com>

and everybody else for their insight, information
and guidance.. doing err is a very good, but a painfull
way to learn :)

Tommi Ripatti
Siemens Finland
CS student at Helsinki University of Technology

---- the original posting ----

To: sun-managers@ra.mcs.anl.gov
cc:
Subject: Sun Ultra 1 / partition messed

Hello managers,

I managed to mess an Ultra 1 ws pretty well.

When the ws was up and running I pressed Stop-A and tried to
run probe-scsi-all.

The prompt told me that I only could do this after the
machine was rebooted and asked me to proceed (y/n?).
I answered yes.

The machine froze complitely. I had to switch the power off in order
to boot it. Now, I got errors from / partition during the routine
boottime fsck and was advised to run fsck manually.

It prompted to "Press CTRL-D to proceed with normal boot"
or issue root passwd to start maintenace procedures.

Booting failed and I rebooted and run "fsck /dev/rdsk/c0t0s0d0"
which generated so many errors that I got bored and run fsck
again, this time with "y" option.

I know, this shouldn't be done.

OK, now the machine doesn't boot at all.

How can I salvage root partition? Do I have to re-install the
whole operating System (Solaris 2.5.1) or does installing
the coredistribution suffice or is there an other way to repair
the system that doesn't involve re-installing the OS?

What are my options? I wouldn't want to install and configure
all the software again. I do have backups of the major software
components though.

All suggestions are welcome,
thanks in advance.

Tommi Ripatti
Siemens Finland
CS student at Helsinki University of Technology