HP SureStore 4/40 - SCSI bus reset

[Dmitry V. Voronenkov]

We use HP SureStore 4/40 with two Ultrium drives. We use Solstice Backup (Legato networker witch Sun label) as backup software. And have two problems:

First:
SCSI bus reset occurred when software try moving tape in library.

Second:
This error occurred while Networker write on disk.
Feb 16 03:37:57 e4000 unix: WARNING: /sbus@2,0/QLGC,isp@2,10000/st@2,0 (st44):
Feb 16 03:37:57 e4000 Error for Command: read Error Level: Fatal
Feb 16 03:37:57 e4000 unix: Requested Block: 0 Error Block: 0
Feb 16 03:37:57 e4000 unix: Vendor: HP Serial Number:
Feb 16 03:37:57 e4000 unix: Sense Key: Media Error
Feb 16 03:37:57 e4000 unix: ASC: 0x14 (recorded entity not found), ASCQ: 0x0, FRU: 0x0


Can you help on solve this problems?

[Alexander M. Ermes]

Hi there.
I think you have four drives, otherwise it would be a 2/40.
Did you daisychain the transport robot and the drives on one scsi interface ?
If ye, change the setup.
one drive --> one interface
the robot can be hokked to one drive, but nothing else.
Check the setup of your robot in your backup software as well.
Rgds
Alexander M. Ermes

[Vincent Farrugia]

Hello,

Check out this link for correct SCSI connection to a 4/40:

http://www.hp.com/cgi-bin/cposupport/get_doc.pl?SNI=hpsurestor19252&LC=information_storage&Tfile=lpg75041

If you have fibre channel connection, check out this link:

http://www.hp.com/cgi-bin/cposupport/get_doc.pl?SNI=hpsurestor19252&LC=information_storage&Tfile=lpg75042

HTH,
Vince

[Dmitry V. Voronenkov]

to Vincent Farrugia:

Our drives attached as showed on Figure 4 in our link (two drives and robot on one SCSI controller). I try connect only one drive and robot, but with same result :(

[Vincent Farrugia]

Hello,

What firmware version do you have on the controller? You require a firmware version of 1.20 or later to support Ultirum drives.

HTH,
Vince

[Dmitry V. Voronenkov]

to Vincent Farrugia:

Library Firmware Rev: 1330
Drives Firmware Revision: E12V

[Vincent Farrugia]

Hello,

Regarding second problem... did you try to replace the tape cartidge with a new one and see what happens? It seems like a faulty tape cartidge.

Vince

[Dmitry V. Voronenkov]

to Alexander M. Ermes:
no, only two drives. Ok, 2/40.

We connect two drives and robot on one SCSI controller. HP allow this as described in

http://www.hp.com/cgi-bin/cposupport/get_doc.pl?SNI=hpsurestor19252&LC=information_storage&Tfile=lpg75041

Figure 4.

I test configuration with one drive and robot on SCSI controller, but with same result.

[Vincent Farrugia]

Hello,

Ah, now I know your config. You are daisy-chaining drives.

I doubt that will work for Ultrium drives though. Ultrium drives require a dedicated SCSI bus.

Try to connect 1 drive using the Figure 1 method, not the Figure 4 method.

HTH,
Vince

[Dmitry V. Voronenkov]

to Vincent Farrugia:

Yes, i try more tapes. As i can see, ca. 50 % tapes work with this error, some work fine(i already doubt aout this), some work with media errors on tape.

I think this is not normal.
Can you say about your tapes quality?

[Dmitry V. Voronenkov]

to Alexander M. Ermes:

I try this at this time. Same result: SCSI reset when Networker loading or ejecting tape and Media errors on tapes.

Can you say about error: "recorded entity not found"?

[Vincent Farrugia]

Hi,

50% 50% is a bad sign, I suspect it's a configuration problem.

As I said, Ultrium drives cannot be daisy-chained. You may work with DLT8000 drives with loss of performance but I doubt Ultrium drives will work at all if you daisy-chain them.

I think the problems you are having relate to the daisy-chaining of your Ultrium drives.

Try to connect 1 drive as shown on Figure 1.

HTH,
Vince

[Frank Abbott_1]

Hi,

One thing I noticed is that the library currently has E12V firmware. This is old firmware, and you might wish to upgrade to E1AV firmware (recently released). If you have access to an NT system or a HP-UX system, the firmware can be downloaded from HP's web site at http://www.hp.com/cgi-bin/cposupport/swindex.pl?p=hpsurestor18482&l=en&c=FirmwareBundle. For the Sun system, you need a FUP tape to upgrade your drives and need to call your local retailer or HP's support at (970) 645-1000.

[Kevin Chappell_1]

the error you are seeing "entity not found" is an indication that your take at some point in time was not closed out properly. This means that during a write of the tape some event occurred, such as a power outage or a power cycle of the drive or better yet a Scsi bus reset. This causes a problem for you now because the cem on the tape cannot be updated with a new "directory". The "directory" allows for fast locates on the tape and without that when you load the tape and attempt to locate to a spot down the tape you end up spacing file by file until it finds it. Needless to say, this can take quite a while depending upon the amount of data on a tape. Try a space to EOD on one of these tapes and you will see the 03 14 00 come up. The other part of this problem is that if this is attached to a HPUX system then the stape driver that you are using has a 10 minute timeout, so this will cause you problems if this file by file spacing takes longer then 10 minutes. Usually you can fix a "bad directory" by doing a space to EOD/rewind/unload which rewrites the directory information, however with the tape not closed out this will never work. Your only option is to copy off all the good files off until the point where this last unclosed out file is and then get the tape reinitialized (not sure how that is done on Ultrium). The other people are probably right about the configuration being the root of your problem. More then one drive per interface opens you up to problems. Related to your problem, what if something occurred on one drive while the other is writing data? Well, a scsi bus reset may happen and if so your backup is finished and your tape is not closed out.

[Dmitry V. Voronenkov]

to Kevin Chappell:

You wrote: "Try a space to EOD on one of these tapes and you will see the 03 14 00 come up." What that mean?
move to EOD: mt eom
What's mean digits 03 14 00?

At this time i disconnect second drive and work with only one, but same error occured.

[Kevin Chappell_1]

I think you have missed the whole point of my explanation. You now have tape media that are the problem and causing you the 03 14 00. Your system configuration may be the ultimate problem that caused the tapes to be this way, but you now have tapes that need to be fixed as I described in my last message. An 03 14 00 is the scsi sense byte return from the problem tape. The 03 is the sense key and means it was a Medium error. The 14 00 is the ASC and ASCQ(Additional Sense Code and Additional Sense Code Qualiflier). In this case, 03 - Mediaum Error, 14 00 - Recorded Entity Not Found.

mt EOD or mt EOM will go to the end of tape, unless it takes longer then 10 minutes in which case the stape driver will timeout. However, in the case with your tape it will never find the EOD filemarks, because the tape was never closed out properly and you will get the 03 14 00.