-
Title
Restoring data from an LTO-3 tape causes the drive to go down with a channel error. -
Description
Restoring data that resides on an LTO-3 tape causes the drive to go down in NetVault Device Management and the job fails with the following error:
Error 2013/03/22 10:25:57 1592 Media DOUGCLUSTB DOUGCLUSTB 4-0.0.0 (HP Ultrium 5-SCSI): had channel error
-
Resolution
Error 2013/03/22 10:25:57 1592 Media DOUGCLUSTB DOUGCLUSTB 4-0.0.0 (HP Ultrium 5-SCSI): had channel error
Sub Text:
TAPESCSI:01976 10 0 102319 Read header from tape (bs = 1024) (269) at 1:0 (END)
SCSI :01976 14 0 102319 CMD MODE SELECT [ 15 00 00 00 0c 00 ]
SCSI :01976 16 0 102319 CMD STATUS 'GOOD' [0]
SCSI :01976 20 0 102319 CMD RETURN VALUE 0
TAPESCSI:01976 57 0 102319 Block size set to 65536
The tape library drive configuration had the Media Block Size (MBS) set to 64, but could see from the above diagnostics that the MBS was set to 1024 at the time of the backup (bs = 1024).
Trying to perform a Data Copy from the old LTO-3 media to LTO-5 also failed with the following errors as well:
1] Error 2013/03/22 10:25:57 1592 Media netvault Command failed, re-initializing driver
With sub text:
MEDREAD :05716 301 0 102429 To block 35812 offset 1048575
MEDREAD :05716 139 0 102429 Request is for more than one block, 1048576 bytes in first block, 1048576 byte in last block
MEDREAD :05716 143 0 102429 The block requested is the next block
MEDREAD :05716 323 0 102429 First 0 0 1048576, Last 35812 1048575 1048576 Cur 0 End 35813
MEDREAD :05716 306 0 102429 DevMgrReadData()
MESSAGE :05716 57 0 102429 MsgReceive(0018F6C4, 00A0CBD0)
MESSAGE :05716 55 0 102429 MsgUpdateReceiveList(00A0CBD0)
LIST :05716 17 0 102429 TRUE = ListWalk(00A0BDC0, 0067AF30, 0257C5D8)
NET :05716 426 0 102429 Selecting for [ ]
NET :05716 38 0 102429 Wait 1851731032 milliseconds for data
NET :05716 40 0 102429 3 = NetSelect(0018F59C, 02564030, 0257C5D8) 0
MESSAGE :05716 227 0 102429 Updated receive list
MESSAGE :05716 58 0 102429 FALSE = MsgReceive(0018F6C4, 00A0CBD0)
DATA :05716 133 0 102429 Found write buffer 0
MEDREAD :05716 340 0 102429 Marking block as first of xfer
CAMNT :05716 34 0 102429 ScsiCamCmd(02591088, 02595A80)
CAMNT :05716 36 0 102429 Build the command
CAMNT :05716 37 0 102429 Issue the command
CAMNT :05716 38 1 102429 (NT) Device I/O control failed (22,87)
CAMNT :05716 40 1 102429 (NT) Windows NT Kernel SCSI Passthrough unrecoverable NT kernel error
CAMNT :05716 41 1 102429 (NT) SPT 000018f304: 2c 00 00 00 09 00 06 18 01 00 00 00 00 00 10 00
CAMNT :05716 41 1 102429 (NT) SPT 000018f314: 84 03 00 00 f8 07 13 0b 30 00 00 00 08 01 00 00
CAMNT :05716 41 1 102429 (NT) SPT 000018f324: 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
CAMNT :05716 42 1 102429 (NT) SCMD 0002595a84: 08 01 00 00 01 00 00 00 00 00 00 00 00 00 00 00
CAMNT :05716 44 0 102429 Command complete
SCSI :05716 15 0 102429 CMD failed: errno 87
2] Error 2013/03/22 10:25:57 1592 Media netvault 'Milla-LTO4-1: DRIVE 2:netvault' has gone down
Changing the MBS on the drives to 1024 didn’t resolve the issue either.
Resolution:
Make sure that the SCSI card can be setup to use a MBS of 1024 (please consult your hardware vendor for the instructions), as well as making sure that the Windows Operating System where the NetVault Server resides is a 32Bit version, as a 64Bit version added to the problem due to the old SCSI card hardware as well. The original NetVault Server was installed on Windows 2003 32Bit and the current NetVault Server resided on Windows 2008 64Bit.
Alternatively you could install a NetVault Client on a 32Bit Windows Operating System and attach the tape library and SCSI card there i.e. becomes a Smart Client (you will need a Smart Client license for this).
We highly recommend that you don’t use a MBS of 1024 as this is way too high for Windows. The maximum should be 256 for a FC connection with Windows, but 64 is a good compromise between backing up small and large files.