My install problem resided in an "ultra" [ie, fast-20, 20 MHz, not wide-scsi]
speed scsi drive.
Ultra-scsi worked, tape drives worked, but an ultra-scsi disk drive and
a scsi tape drive would not work on the same daisy chain.
On bootup, I entered my Adaptec 2940UW setup to disable ultra-scsi speed.
Below I give the history of my problem with some error messages I saw.
I have two disk drives, one ultra-scsi and one fast-scsi,
which Debian uses well.
Last month, I bought a used Exabyte 8200 tape drive for $160.
My problems apparently started when I attached a tape drive (any tape drive).
I made a perfect backup on tape, so the tape drive worked.
But a week later, I finally (should have seen earlier) noticed some bad files
on my Debian distribution.
I tried to correct the problems through massive dpkg installs and reinstalls.
But after two weeks more commands were failing, so I decided to upgrade the
package "dpkg" to Debian 1.3.
I had used without problem numerous other Debian 1.3 packages for months,
so I expected little problem upgrading "dpkg" and I did not look at any
readme's.
Of course, this naive upgrade added even more problems to my already numerous
bad binaries.
I proceeded to install Debian 1.3 from scatch.
Installation of Debian 1.3 from scatch produced errors.
This combined with the dpkg problems I had been having for two weeks,
led to my conclusion that I miss-interpreted how to install Debian.
I then sought help from the Debian community.
Some of the errors I saw I list below.
Some of these are very weird, reflecting binary corruption.
1. Installation of Debian 1.3 left first line of /etc/lilo.conf with
boot=
instead of
boot=/dev/sda2
During installation, when I entered <OK> for
Next Make Linux Bootable Directly From Hard Disk
I might get
Currently it is impossible to boot from the second harddisk.
Please boot the system using the rescue boot method
and configure LILO manually.
2. During Debian 1.3 installation, by hand I altered /etc/lilo.conf with
boot=/dev/sda
then entered
/target/sbin/lilo -r /target
This seemed to work, but rebooting I got the error
.
.
.
LILO loading Linux
Uncompressing Linux...
crs error
--System halted
3. I could almost get a floppy boot-record to work, since
the installation indicated no problems.
On booting from floppy, the boot stream looked perfect,
but hung eventually after giving the line
VFS: Mounted root (ext2 filesystem) readonly.
4. During install, after configuring the network, the installation routine
cycled back to setting color (the very first step of installation).
5. After installation, I could boot, but after login I could not interract and
within about 5 seconds I was given the login prompt again.
Amidst all my problems, I had long ago turned off my Exabyte tape drive
residing at the end of my scsi daisy chain, EXPECTING THE SCSI DAISY CHAIN
WOULD ACT AS IF I HAD TERMINATED AT THE LAST DISK DRIVE.
I NOW KNOW THIS IS NOT TRUE.
Removing the tape drive, Debian installed cleanly.
So, I stopped suspecting Debian and started suspecting Exabyte.
I substituted a DDS-3 4mm tape drive, but got the same problems that I got
with the Exabyte drive.
So, my problem appeared to be tape drives in general.
With either tape drive attached and turned on, Debian 1.3 installation would
fail.
As I altered my computer system elsewhere, I would check Debian's initialization
of swap space, then its initialization of a Linux partition.
These both usually failed when I had the tape drive attached.
If I got a seemingly clean Debian 1.3 installation,
with the tape drive attached, on booting I would get the following errors.
6.
scsi : aborting command due to timeout: pid 61; scsi 0, channel 0, id0,
lun 0 Read (6) 02 38 58 78 00
aic7xxx: (abort) Aborting scb 0, TCL 0/0/0
7.
INIT: Entering runlevel: 2
INIT: ID "1" responding too fast:
disabled for 5 minutes.
every five minutes.
I tried making my scsi cables strait, though this required that
I have the geometry
---------- -------
|computer| --------------------------------------------- |ultra|
---------- ------------ |scsi |
|tape drive| ----- |disk |
------------ -------
This straight geometry had two scsi cables somewhat in parallel,
creating even more error messages.
I would guess that one wants cables perpendicular, not in parallel.
And I would guess that scsi cable twists matter little so long as one does not
form 360 degree curves.
I tried APS Technologies SCSI Sentry, a digital active terminator.
This removed the installation errors when initializing swap space or
initializing a linux partition.
Since I happened to have two spare 32MB partitions,
I did separate installs on both.
The installs went flawlessly, but WHEN I COMPARED (diff -r) /bin AND /usr/bin,
SOME FILES DIFFERED. These files had the same byte size.
I wonder what this digital filter did.
Perhaps it managed to hide errors that the usual active terminator
would not hide.
ALL THESE PROBLEMS DISAPPEARED WHEN, on bootup, I entered <control>-a,
then altered the Adaptec 2940UW scsi setup.
I DISABLED "support for ultra speed".
PS: Debian 1.3 installation now works perfectly for me by 7 floppy disks
or by 1 floppy disk and a mounted mirror of Debian.
I use some of the improvements on the previous Debian distributions.
I thank Bruce Perens and Nathan Norman for their help and tricks; eg,
lilo -r /target
Jim Burt, NJ9L, [EMAIL PROTECTED]
--
TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
[EMAIL PROTECTED] .
Trouble? e-mail to [EMAIL PROTECTED] .
