System can not boot

[admin1979]

Hello,

Today we noticed the one of the Alpha Server TRU64 is at the boot prompt. The boot prompt
is P000> something. We tried the commands boot but the error says ,

Failed to open dka0.0.0.1.0 . What do we do please suggest.

Thanks,
admin

[Steven Schweda]

> [...] the Alpha Server [...]

Describe?

> We tried the commands boot [...]

Actual command with actual output?

>>>show device
>>>show boot*

[admin1979]

I apologize.

> [...] the Alpha Server [...]

Its Alpha Server 2100 running TRU64 V4.0G.


> We tried the commands boot [...]


>>>boot

Failed to open dka0.0.0.1.0

>>>show device

dka200.2.0.1.0 DKA200 RZ28 442D
dka600.6.0.1.0 DKA600 RRD43 1084
dva0.0.0.1000.0 DVA0 RX26/RX23
mka0.0.0.1.0 MKA0 TLZ9 A028
ewa0.0.0.0 EWA0 08-00-2B-EZ-7A-58
pk00.7.0.1.0 PKA0 SCSI BUS ID 7
pk00.8.0.1.0 PKA0 SCSI BUS ID 7 F01 A12

>>>show boot*

boot_dev dka0.0.0.1.0
boot_file genvmunix
boot_osflags A
boot_reset OFF
bootdef_dev dka0.0.0.1.0
booted_dev
booted_file
booted_osflags


Plz let me know if you need any other info.

Thanx
admin

>>>show boot*


Tried command ,

>>>boot
Failed to open dka0.0.0.1.0

[Steven Schweda]

> [...]
> >>>show device
>
> dka200.2.0.1.0 DKA200 RZ28 442D
> dka600.6.0.1.0 DKA600 RRD43 1084
> [...]

dka0 seems to be missing/dead.

Cycle the power? Check the cables? Replace
the disk? Undo whatever have you done lately
to wreck everything?

[admin1979]

I followed what you said and recycled the system. Yesterday one of the admins connected a Alpha server backup tape library.
And suddenly the system went to P000>> prompt. When we noticed this, we powered off the tape and recycled and dka0 comes back.
Able to boot the system as well.
Now the question, why it happened? The tape was connected to the scsi port available at the back-plane of Alpha 2100. Is it not the right place to connect the tape? I doubt.
Something is preventing the root disk and tape library to be able to function simultaneously.

Moreover , to debug the issue I replaced the root disk onto another free slot but then it was detected as dka100 instead of dka0.
Then to I gave boot dka100 and the system tried to boot but hung at INIT spawning to fast.

I guess since it could not find the original dka0 disk at that place. Or what?

[Martin Moore]

Here's what probably happened...

Your tape device is SCSI target 0 (shown by the device name MKA0), which conflicted with the boot disk when the disk was in the original slot (DKA0), preventing the disk from being seen.

When you moved the disk to another slot, that caused it to be seen at target 1 (dka100). But in Tru64 UNIX version 4, device names are based on SCSI bus/target/lun. When the disk is at DKA0, the device name is rz0, but at DKA100, it's rz1. However, all your file specs in /etc/fstab, /etc/fdmns, and /sbin/swapdefault will still point to rz0 -- which doesn't exist when the disk is DKA100 rather than DKA0. At boot time, the system wouldn't be able to mount /usr (assuming it's on the boot disk), which leads to the "respawning too rapidly" message.

Easiest solution: Leave the boot disk in its original slot. Change the tape device to have a different target ID that's not currently used on that bus, such as 3. You'd have to look at the tape drive documentation for instructions on how to do that. Then plug the tape in and it shouldn't conflict with anything.

Alternative: Leave the tape drive alone, but edit the relevant files on the system disk to change rz0 references to rz1. Then move the disk to the other slot you tried, and you should be able to boot it as dka100.

Martin

[Steven Schweda]

> Yesterday one of the admins connected a
> Alpha server backup tape library. [...]

And you didn't think that this was important
until now?

> Something is preventing the root disk and
> tape library to be able to function
> simultaneously.

Cable too long? Unwise choice of SCSI ID on
the tape drive? Before you touch the
hardware (again), you should do some reading
on SCSI configuration. Playing with a SCSI
bus on a running system can often be done,
but playing with the SCSI bus with the
system/root/boot disk on it (while the system
is running) is asking for trouble.

First: Every SCSI device has a SCSI ID, and
you can't have two devices with the same SCSI
ID on the same SCSI bus. If you don't know
which IDs are in use on the bus, and you
don't know the ID of the device you're about
to connect to the bus, then you shouldn't be
allowed to touch the hardware.

> I guess since it could not find the
> original dka0 disk at that place.

Probably a good guess. It might work with
VMS, but most UNIX(-like) systems have
physical device (file) names in their
"/etc/[v]fstab" files (or somewhere), so you
can't simply change the name of the disk and
expect things to work.

By the way, some tape libraries use more than
one SCSI ID. You may need to read the manual
to learn how to configure the thing.

Also, while I'm complaining, "backup tape
library" is not a very complete description
of the device in question.

[admin1979]

Yes Martin , you hit the deck. Will try out the alternatives you have suggested.
If the 1st option does not seem to work due to lack of docs, in order to try out second option , how to edit the relevant files on the system disk to change rz0 references to rz1.

As a trial , I did boot the system thru TRU64 Instalaltion CD media by giving,
P000>>>boot mka0
I have been told that we need to go to the shell prompt for diagnosis but I did not happen to see that option. The system booted and the installation started with the questions like Installation Language : English and so.

Am I missing anything out here?

[admin1979]

I understand Steven. I do lack knowledge of SCSI devices that I would try n refresh.
But since the system in questions is one of the development systems hence we could dare to do R&D on this system. But in the end, we need to bring this system up.I hope you understand now.

[Rob Leadbeater]

Hi,

> P000>>>boot mka0

If that is what you typed then you booted up off the tape drive, not the CD ROM drive which may well explain why didn't see the "shell option".

Your CD is on DKA600, so

P00>>> boot dka600

As you're obviously new to Tru64 I would strongly suggest getting hold of a copy of PersonalAlpha so you can play around with things....

http://www.emulatorsinternational.com/en/personalalpha.htm

Cheers,

Rob