Thankyou for all the helpful replies. Sorry for the delay in
summarizing. Original post at the end.
I haven't had to try most of these solutions yet. My original problem
turned out thankfully to be just a loose SCSI connector from the
station to an external disk. It looks like I'll get to try the others
soon enough though :-( Now a second station is coming up with 'corrupt
label - bad geometry' messages, and a third station recently had the
Quantum "sticktion" problem on both internal 105 MB drives when taken
down for scheduled maintenance.
=============== SUMMARY of replies ===================================
(1)General SCSI/hardware scenario:
Check the easy things first. Any problem with the SCSI chain on this
workstation? Improper termination, loose connection or one of the
SCSI devices had power fail? Make sure everything is seated properly,
the scsi connectors are tightly connected, the terminator didn't fall
off, the drive is getting power, the fan spins, etc. Is the drive
making any wierd noises?
(2)"Sticktion" problem:
Quantum 105MB drives are notorious for this. If you power down the
station during reboot, you may need to do a few tricks to get it
started again. Remove the drive (anti-static precautions taken) and
give it a sharp twist or hit it with the heel of your hand or a
screwdriver. Sometimes you have to hit the drive fairly hard to get
the heads freed up.
(3) Boot block corrupted or Disk label gone bad:
Boot off the cdrom (or server over the network or another bootable
disk), and then try one of:
(a.) 'fsck' the boot partition. May need to use an alternate
super block (see "man fsck"). 'fsck -b' will search for a backup
label and install it as the primary one.
(b.) 'format' has an option to restore the primary from the
backup label. The label includes the disk geometry and the partition
table. Use the 'analyze' function of the the format program to
non-destructively analyze the boot disk. Be sure and read the
appropriate sections of the Network and Administration manual --
format can completely wipe out your disk if misused! If format cannot
analyze your disk, you probably have a hardware problem.
(c.) run installboot(8) from the miniroot.
(d.) try to install a new boot block using MAKEDEV.
(4) Unique:
From: Birger.Wathne@vest.sdata.no (Birger A. Wathne)
If it's a SS2 with an early Wren VII disk, it could be that he had a
patch in NVRAM for this disk. This patch may have been erased somehow.
The patch would be in the nvramrc variable in the boot prom. If you
have lost the patch (and this is a WREN VII disk) I may be able to
find it for you....
----------------- ORIGINAL POST --------------------------------------
Date: Wed, 16 Jun 93 12:15:06 EDT
From: rcc@next1.math.pitt.edu (Robert C Cyphers)
To: sun-managers@eecs.nwu.edu
Subject: Bad disk Label (can't boot)
Cc: rcc@next1.math.pitt.edu
Status: R
Hello Sun Managers,
One of my users (Sparc 2 SunOS 4.1.1) has just had his workstation
blank out on him while he was using it. Attempting to reboot it gave
the following error message:
>b
Boot device: /sbus/esp@0,800000/sd@3,0 File and args:
Bad magic number in disk label
Can't open Sun disk label package
Can't open boot device
>
Any idea what's going on and/or how to fix it?
Robert C Cyphers
Dept of Mathematics and Statistics
Univ of Pittsburgh
cyphers+@pitt.edu
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:07:58 CDT