Problems with IBM 3485 Library with Fiber LTO Ultrium Drive

[Striker9]

Hey Everyone

We are currently configured with and Sun Networker Server (6.02) connected via SCSI Fiber through a fiber channel switch to and IBM 3485 Library with 6 LTO Ultrium tape drives.

We are experiencing drive failures at least once a week and working with IBM we found that the drives appeared to be receiving the SCSI commands in the wrong order. IE Eject tape would come before a write or read to that same device.

Wondering if there is anyone else out there run with LTO Ultrium Drives or other fiber connected devices that are or did experience simular problems and what you did to get around them.

Stephen

 

[1407]

Are you using smartmedia ?

 

[Striker9]

We are not running Smart Media today. We were going to be moving to that in early next year but with AlphaStor replacing it we have held off on the Smart Media and are looking to move to AlphaStor.

 

[jpj]

We have just implemented a backup/restore solution based on sun backup server, Legato 6.1.1 and also fibre switch connected to 5 X SCSI IBM LTO 3580 drives. The fibre is terminated with storagetek routers (L700 library) and then scsi output to the drives.
Its early days for us and I am new to Legato but we have been experiancing some problems with drive errors. The logs read like the following;
"/dev/rmt/4cbn: i/o error
Then says tape is full (physically not) block size is 32768 not 65536 bytes.
verify the device configuration - tape positioning by record is disabled."
This will happen to two or three drives at once.
What generally happens then is that the drive will attempt to load other same pool tapes to carry on with the backup but the same series of errors occur.
Eventually the drives are disabled - probably correct.
Following this other drives (max 5) will start to give the same errors until all drives become disabled.
The tape drives were giving errors on tape library display as "not connected" on some but not on others.
Storagetek have recently upgraded the firmware and have not seen these errors since.
We are also using Storagetek ACSLS software to control the library.
It would be good to know if you found a fix for your drive errors and may be a clue as to the problem we are experiancing.

 

[Sunboxez]

Hey jpj,

I have an STKL40 with 3LTO Ultrium Drives on a NT Network and before we went to the 1400 firmware revision we had the problem of tapes being marked full prematurely. Since your problem has been resolved, have you tried restoring data from any of your tapes that were written/labeled/formatted while the problem was occuring? When we tried to restore data from those tapes we recieved error:unable to postion media to file x, record y.

Tapes that were labeled during the bad firmware will have to be erased via Ntbackup or a different backup software to fully erase the networker labels that were written to tape. Simply recycling or relabeling tapes in legato left us with the wrong blocksize, so our tapes still had errors.

In my case that firmware issue seemed to have changed the fixed blocksize to 32K or 32768 Networker filemarks being written to the tape expected 64K or 65536, hence the filemarks didn't match and legato thought the tape was full. If you run the mt -f <device> stat | more command you should see your drive has a default blocksize of 32768 and a max of 65536. If your <device> is NT you use \\.\TapeX if Unix /dev/rmt/Xcbn check your docs.

If you decide to try any of this to find what caused your problems please let me know.

Thanks,

Larry

 

[Striker9]

It appears that a firmware upgrade for us on both the drives and the IBM 3584 library has helped us out as well.

Thanks for that tip Larry.. Will check those tapes now and see if we have the bad block problem..

Stephen

 

[jpj]

Larry - we are carrying out an new implementation of a backup system, &quot;I am on a steep learning curve&quot; and also we have more than one problem. I may have misled slightly above as when the firmware on the drives was changed, &quot;Storagetek have recently upgraded the firmware and have not seen these errors since.&quot; What I should have said was that the errors had changed, which is probably reasonable to expect, but we had the same problem of the drives becoming disabled. So far one of our problems was with the ACSLS library control software which has now been patched, meaning that the library will remain controllable when the drives start to play up. Before we needed to restart ACSLS before we could even manually change a tape. By the way, after Legato disables a drive we need to manually load a tape using ACSLS, and then Legato will unmount it even though it shows an unmounted status!!
Our problem always starts with drive 3stcbn therefore its either a faulty drive or I am beginning to look into load and unload sleep times. What is happening now is that we have 5 drives split into 2 X Storagenet routers, 1 X 3 and 1 X 2. All drives on the 1 X 3 leg including 3stcbn get disabled in Legato during the fault condition. Our fault condition only occurs during full backups, and we do not experiance any problems when this drive is manually disabled. Also the fault condition always happens when the tape is deemed full and a tape change is requested.
I am going to up the sleep times and see if this helps.
All suggestions are appreciated.