External DLT8000 removed-reconnected and not seen (AS4100 w/Tru64 5.1b)

[Horacio Rivero]

Hello dear friends, I have this problem:

An external DLT-8000 was removed with the SCSI cable for repairing from an AS4100 with Tru64 5.1b. The system was reinitiated without tape drive, then the tape drive and SCSI cable were re-connected but not seen in "hwmgr -v dev". Also it is shown as null with "hwmgr -show scsi".
There are also SCSI errors and bus resets in messages file.
In the other side, is it possible to damage the SCSI controller when connecting cable and drive while system is working?

I will apreciate all your help.

Please tell me if you need more information.

Thank you

Regards

[Steven Schweda · Accepted solution]

> [...] is it possible [...]

Many things are possible, and many are more
likely than damaging the SCSI adapter, but
you _could_ do that. I'd vote for simpler
SCSI problems, like bad cables or
termination, or ID conflicts.

What are all the things on this SCSI bus (at
which IDs)? What does hwmgr say if you
disconnect tape drive again?

I assume that you were "connecting cable and
drive while system is working" because you
didn't wish to shut the system down, but it
might be interesting to see what the SRM
console says to "show device".

[Rob Leadbeater]

Hi,

Did you do a "hwmgr scan scsi" after reconnecting the drive ?

Was the DLT drive actually replaced, or just repaired...?

Cheers,

Rob

[Vladimir Fabecic]

If tape drive was replaced, it will not be named tape0 but tape1 (or simular).
Please do:
# hwmgr -scan scsi
and send output of:
# hwmgr -show scsi

[Horacio Rivero]

Hi friends,

Steven,
the only device on the bus is the tape drive. In fact yes, the system was not available for shutdown in that moment (Oct 5th). Then, the system was available (Oct 10th) and what I saw is that in SRM console, the tape drive was not shown so, I did take out the controller, the cable, and tape unit for testing in other system and the failure is the controller. I am worried because I replaced the controller few months ago because of a similar failure and now this controller has been damaged again, what make me think that something is damaging the controllers.

Rob,
The internal drive of the unit was replaced. I did the scan scsi after reconnecting but no results.

Vladimir,
I will replace the controller and will do what you recommend.

Thank you very much to all of you. I will give you my feedback.

Regards.

[Rob Leadbeater]

Hi,

What model controller is it ?

You're not connecting a differential (HVD) device to a single ended (LVD) card are you ?

You might also want to check that there are no bent pins on your SCSI cable.

Cheers,

Rob

[Horacio Rivero]

Hello,
I'm sorry dear friends, I was out.

Rob,
I checked the pins and they are ok. The SCSI controller is SN-KZPCA-AX and DLT unit is SE.
I checked the DLT-8000 unit, SCSI cable and 2 SCSI controllers (SN-KZPCA-AX) in other system, verifying that the problem are the SCSI controllers.
I'm worried because 2 controllers have been damaged and if I connect a new controller in the production system, it will be damaged again.

I was thinking about the voltages given by the PCI slots (AS4100) and the used by the SCSI controller. Do you think this the reason for the controllers to be damaged? I verified and PCI slots use 5v but I don't know about KZPCA.
I was also thinking about if there is a bridge between neutral and ground (in outlet)that is causing some noise affecting the controller, what do you think?

Regards

[Rob Leadbeater]

Hi,

According to the options list the KZPCA is supported on the 4100, so that shouldn't be the problem.

http://h18002.www1.hp.com/alphaserver/options/as4100/as4100_15_73_options.html#OptListHead

How exactly have you determined that it is the SCSI card that is faulty ?

I've still got a feeling that you may have a problem with the tape drive... Have you verified that the unit inside the external box is definitely single ended. I know from experience that the parts that get sent out are not necessarily what they should be !

Cheers,

Rob

[Steven Schweda]

> I verified and PCI slots use 5v but [...]

As I understand it, PCI slots are keyed
according to the voltage, so this shouldn't
be a problem. For example, see page 7 of:
http://www.digi.com/pdf/prd_msc_pcitech.pdf

> [...] bridge between neutral and ground [...]

Possible, I suppose. My peripherals are
normally powered from the same outlet as the
system, which makes it hard to get that kind
of problem. You could try measuring the
voltage between cabinets to see if it's not
very close to zero. (Or if you see sparks
when attaching the cables...) If you're
really frying adapter cards at this rate,
then a bad ground may be a good theory.

[Horacio Rivero]

Hello

Rob,
In order to test, I connected the tape unit on an Alpha Server 1200, of course with the controllers and they don't show the unit in SRM console (show device), even the controllers are shown as "unknown". Then I tested with a KZPBA-CX controller and DLT unit is shown with "show device" (I didn't test w/O.S.), but after this, I installed a new KZPCA controller in this same system and the DLT unit is shown in SRM console (show device) and even with Operating System (Tru64 5.1b). I did some vdump tests and worked fine. That is how I determined the controllers are faulty.

Steven,
I will test the outlets and will verify if there is a difference between the outlet of the system and the outlet of the DLT unit.

Any other suggestions please let me know.

Thank you for your help and will give you my feedback.

Regards.