Problem with FC Attached SureStore 2/20

[Duncan Edmonstone]

... Well I finaly got back to looking atthis problem - thanks Vincent the patch you suggested did look like a likely solution, but no it didn't work.... We';ve also tried the same unit attached to a Win2K machine and had very similar problems (the two tape devices were fine, but the robotics report as 'OFFLINE' and no way to bring it online).

Looks like its time to log a hardware support call!

Cheers

Duncan

[Curtis Ballard]

Have you checked the RMC (network remote management card)? If you open the web pages on your RMC you should be able to see the true library status and narrow down your problem. The RMC status should show something like "checking inventory" if the library really does think it isn't fully powered up. It should show something like "Status: Good" if the library is actually ready.

If the RMC reports Good status then the possible problem locations are narrowed down to the the Fibre interface card, or something between it and your host. I don't believe mc uses any cached data so the problem probably isn't the host.

If the RMC doesn't report status Good or if you can't access it, the most common cause of a library not reporting good status is an unlocked or damaged drawer. Try opening all of them and if any opens then it needs closed and locked. If none of them open but the RMC or front panel reports "Door open" then a sensor on one of the drawers may be damaged and a hardware support call will be necessary.

[Sanjay_6]

Hi Duncan,

Try this.

To check the status of library,

mc -p /dev/rac/c20t12d0 -r DS

to move a DLT from slot 1 to Drive 1

mc -p /dev/rac/c20t12d0 -s S1 -d D1

To move the DLT from drive 1 to slot 1,

mc -p /dev/rac/c20t12d0 -s D1 -d S1

Hope this helps.

Regds

[Vincent Farrugia]

Hello,

Please keep us informed on how things develop. It's good to learn something new.

Thanks,
Vince

[Duncan Edmonstone]

Well I do like to post my solutions, even when it takes me six months to arrive at them!

After working on an entirely seperate assignment for the last few months, I'd forgotten all about this problem, right up until the point I had to arrange a demo which used this fibre attached 2/20 - I'd assumed that one of my colleagues had gotten HP in and fixed the issue ages ago, but as it turns out the 2/20 has sat in a corner 'broken' for ages!

After several hours reminding myself everything I had done, we logged a hardware call, and got some HP CEs in to take a look...

After much to-ing and fro-ing and replacing almost every component in the 2/20 several times, we finaly hit on the problem...

The clue to the problem was actually in my original post:

"I am on the latest firmware release, downloaded and installed from HPs web site just yesterday"

Previous to the firmware upgrade I was on version 1.14 of the firmware, and I upgraded to version 1.33.

It turns out that there are some internal interfaces of the 2/20 known as 'bridging links' - these manage the interfaces between the cards (fibre controller/remote management card/library controller) - the firmware upgrade from 1.14 to 1.2? (something) made some changes to these bridging link interfaces, and guess what? the 1.33 version assumes these changes have already been made. The upshot of all this, is that upgrading straight from 1.14 to 1.33 appears to work 100% successfully, but completely breaks the library... this is made even more complex by the fact that:

a) Certain 2/20s only work with certain fibre controllers... you have to check the serial number of your unit before replacing...
b) you cannot go above a certain firmware rev (1.22 I think) if you have a particular transport mech which only has a 4K buffer (newer mechs have an 8k buffer)

Anyway the fix was to downgrade the firmware down to 1.14 (via the SCSI interface, patching firmware via the RMC or via fibre is apparently unreliable even without all these other problems!), and then upgrade in steps back up to 1.30 (we stopped there as I was out of time).

Lessons learned? well I already knew the golden rule of HP tape libraries, and I can't believe I broke it myself:

"NEVER NEVER NEVER upgrade tape library firmware yourself!!! ALWAYS get an HP CE in to do it"

and if you don't have HP support?

"NEVER NEVER NEVER upgrade tape library firmware AT ALL!!"

Anyway thanks for all the help guys, as usual I learned plenty of other neat stuff while fixing this.

Cheers

Duncan

[Michael Tully]

Hi Duncan,

Thanks for posting the solution. I should have hit on it myself. In the past I've only ever installed firmware on servers, and should have suggested a CE do it. We had a CE do our 20/700 tape library and they blew up the system board. So I guess it can happen to anyone, whether it's the right version or not. Firmware is even more finicky than normal software patches.

Anyway, I'm glad you found the solution.

Cheers
~Michael~