Tru64 and defective disks

[Edward hebert]

We have 2 Tru64 servers in a cluster arrangement, and there are bad hard disks in a couple of StorageWorks 4314 Racks that are attached to the servers (The lights on the hard drives are flashing). Using the hwmgr command, I get a list of all the disks, but being new to both fibre channel as well as tru64, how do I correlate this with the actual physical drives? I tried using the "hwmgr locate" command on a few different drives, but I suspect that this does not work with fibre channel (just scsi), because I can not get the LEDs to blink on any of the drives.
Is there another way to make the drives blink? None of the equipment has hardware path labels, so I can not trace the cables around to determine any paths like I do on non-fibre channel HP9000 equipment. Am I going to have to get into the HSG80 controller to determine this somehow?

[Hein van den Heuvel · Accepted solution]

Edward, Welcome to the HP ITRC Tru64 Forum.

When I get blinking disks, I often try a quick powercycle + re-seat first, hoping that will fix it.

The 4314 is just a container. And if it was connected directly to a straightforwards scsi adapter like a kzpba the locate probably would have worked.
The missing information information here is probably a smart controller like a 'SmartArray' or a kzpcc. Please attach the 'hwmgr show scsi' or 'hwmgr show fibre' and such information in a .txt file
They would map a scsi lun onto one or more disk in that container.
One simple 'trick' I use to locate disk is to light the busy light: dd if=/dev/rdsk/... of=/dev/nul. Just try and see which light come up! Of course that will only work for more or less working disks!

Hope this helps some,
Hein van den Heuvel

[Edward hebert]

Hein van den Heuvel,

Thanks for the information about the dd command trick. I have submitted the results of the hwmgr commands and put them in a single text file. These are the results from one of the servers in the cluster; the other server just has a couple extra drives listed.

Ed

[Hein van den Heuvel]

>> Am I going to have to get into the HSG80 controller to determine this somehow?

Yes. You need to ask the HSG about the disks it owns. It knows how the drives are mapped from the LUn presented to Tru64 to the DISKxxxx devices on the back bus. It can do the locate. It may have placed some disks into 'failedset' and replaced from 'spareset'.

I did not take good notice of that line on first reading. The hwmgr output makes sense with that. So you have a kgpsa (fibre) controller to a hsg80 (raid) connecting to disks. This is consistent with the B/L/T pictures of using the sub-luns.
It is a little surprising to see just 1 path, but that could be as designed. (poorly designed as it gives no protection?). No fibre switch/hub huh?

Connect to the HSG directly using the terminal line, or using cliscript or hszterm (/usr/lbin) to the scp device. Or of course using the storage management console.

Sounds to me you may have some reading of manauls/manpages/help test to catch up with!Or maybe it is time to call support or a consulting organization to help you get more familiar with the setup.

Regards,
Hein van den Heuvel
HvdH Performance Consulting.

[Edward hebert]

Hein van den Heuvel,

You are absolutely correct, I have a lot of reading to do. I actually started going through the manuals before I emailed you. I have only worked with HP, Sun, Linux and SCO systems.
Thanks again for the help.....I really needed to replace those bad drives immediately.
You can consider this forum item as closed.


Ed

[Edward hebert]

I now know the basics of getting access to the drive information.