Garbage entries in "ioscan" with ADIC tape library

[Ollie R]

Hi All,

I have an ADIC Scalar 10K containing 3 IBM LTO drives and connected through a brocade switch to my server. Server is N-Class running 11i Enterprise.

In the "ioscan" I get 3 LTO drives listed at target "0", LUN's "2", "3" and "4" on the correct fc - this is normal - but also entries for the targets which show junk in the "Description" field.

I've included a copy of the relevant section for your reference.

Has anyone seen this behaviour before? I'm stumped!

Thanks,


Ollie R

[harry d brown jr]

Ollie,

Can you post the output of this:

ioscan -fn

I'm not sure what options you used on the ioscan you posted.

Have you had any issues with those devices?


Also, what was the last major patch bundle installed on this machine? If it's not up-to-date, I'd suggest you do that immediately - I personally find the custom patch manager to be a great tool to bring a machine totally up-to-date.

live free or die
harry

[Ollie R]

Hi Harry,

The file I included was an extract of "ioscan -fn" with the irrelevant info removed.

The tape drives seem to work OK. The main problem is the messy "ioscan" which, in itself I guess, is not really an issue for system operation but it is untidy and affects some of our system diagnosis tools. We will also need to connect other systems to the ADIC library and I'm expecting the same results.

Latest patching is May 2002. This is a very controlled environment so we can't just throw an updated patch bundle on unfortunately. I've had a good look through the patches database in any case but can't find any reference to this type of issue in there.

Thanks for your help - any further suggestions would be greatly appreciated.

Cheers,


Ollie R

[Eugeny Brychkov]

Garbage appears when reporting phantom device with Target Id=2. This is reported by the Library. Please check in another operating systems (or another HPUX connected hosts) if it's reported correctly. If yes, then it's problem with Library, but if not, and another HPUX host reports ok then problem with this host where do you see this garbage.
BTW, do you see garbage ALWAYS on the Tgt2? Is it intermittent/jumping from Tgt to Tgt?
If it's possible to check inquiry strings from Library's control panel please do so
Eugeny

[Ollie R]

Hi Eugeny,

It's not just target 2 with the garbage - the "12822" entries are also junk.

We will be testing the config on another system in the next few days to narrow it down.

Thanks for the suggestions,


Ollie R

[Eugeny Brychkov]

Got it. This string is just what library returns for inquiry command for phantom LUN. Try diskinfo'ing these LUNs - you'll see size=0. BTW, are you sure this library supported with HPUX (not drives, but library)?
But, of course, two questions - 'supported' and 'will work' are different :o)
Eugeny

[harry d brown jr]

Check on firmware upgrades for the Brocade switch and the ADIC library. Those values are probably coming from the devices.

live free or die
harry

[Ollie R]

Hi Guys,

If I try to do a diskinfo on one of the devices I get:
$ diskinfo /dev/rscsi/c4t6d0
diskinfo: class ctl not recognized

The ADIC library, IBM LTO drives and switches all have the latest versions according to our representatives at HP and ADIC.

Cheers,


Ollie R

[John Payne_2]

This sounds like some kind of funny zoning issue.

Are you zoning out the drives? Is there anything else on this Brocade that might be presenting itself?

John

[Eugeny Brychkov]

Yes, sorry, diskinfo will not work because these LUNs are classified as SCSI controllers. I think you can see something in STM information for them
If there's brocade switch could you please attach 'supportshow' output?
Eugeny

[Michael Lampi]

It looks suspiciously like the 10K's robot's SCSI ID is set to the same address as the SCSI bus controller; i.e., the same ID on the SCSI bus as the SCSI-to-fibre bridge.

This would cause the robot to appear at each and every SCSI address.

The reason it does not appear at SCSI ID's 2, 3 and 4 is that it responds slower than the tape devices at those addresses.

[Brian M Rawlings]

Ollie: you might narrow this down by disconnecting all of the tape devices from the Pathlight 5000 (SNC?) FC-SCSI bridge box (terminate all SCSI ports with the terminators from the drives), leaving only the robot (library) connected via SCSI. Try the IOSCAN that way, and see if the garbage entries still show up.

Then, terminate the connection to the library/robot, and move the cable to one of the individual drives. If the library or its SCSI ID is at fault here, you should see a clean IOSCAN that shows just one FC connected tape drive.

I don't know if this will buy you anything, but in troubleshooting, simplify. Lowest possible number of elements in the config, never change more than one thing at a time, etc, etc.

Good Luck, and please advise us all once you find the cause, this one could come up again... we may be doing something just like it in the near future.

Best Regards, --bmr

[Ollie R]

Hello All,

Finally the mystery of the garbage ADIC entries is fixed. The solution? A microcode upgrade on the SNCs.

For your further information, in the "ioscan", all target-0's are visible for all hardware paths to the SNC's, very much like with the HP XP disk arrays. The correct description of these "reserved" devices is "ADIC P5000 Invl".

Thanks very much for all your time and help - I've assigned points to you all because you came up with some great ideas!

Cheers,


Ollie.

[Ollie R]

P.S. to Harry d Brown Jr.
If you stick another reply to the end of this page I'll award you 10 points - you were spot on. Sadly, ADIC didn't release this latest microcode until mid January!