Device Timeouts after Upgrade from V12 to V12.5

[Hopper]

After upgrading BE from V12 to V12.5, BE generates 6 timeout events in the Windows 2003 system event log while attempting to inventory our HP DAT72x6 tape library (one timeout for each slot).  The firmware is update to date, all windows patches are up to date, we are using the Symantec drivers for this device.  We have never recieved this timeouts with BE versions 11.x or prior versions of 12.x.  This appears to be a bug or impatient logic waiting for the library to load tapes to the drive.  Other than these errors during inventory, are backups are running fine and not generating similar timeout errors.

Here is the text of the Windows Event log entry:

Event Type:        ErrorEvent Source:    adpu160mEvent Category:                NoneEvent ID:              9Date:                     10/22/2008Time:                     6:02:56 PMUser:                     N/AComputer:          NOTTDescription:The device, \Device\Scsi\adpu160m1, did not respond within the timeout period. For more information, see Help and Support Center at http://go.microsoft.com/fwlink/events.asp.Data:0000: 0f 00 10 00 01 00 6a 00   ......j.0008: 00 00 00 00 09 00 04 c0   .......À0010: 01 01 00 50 00 00 00 00   ...P....0018: 01 00 00 00 00 00 00 00   ........0020: 00 00 00 00 00 00 00 00   ........0028: 00 00 00 00 05 00 00 00   ........0030: 00 00 00 00 07 00 00 00   ........

[Anup]

Event ID 9: Signifies SCSI bus timeouts. These errors usually occur when the SCSI bus resets itself, or the SCSI hardware does not respond in a timely fashion. Sometimes slowing the SCSI bus down can resolve these issues, as well as installing the latest SCSI drivers and firmware. In certain high end servers, moving the SCSI card to a different Peripheral Component Interconnect (PCI) slot, one that does not share the same bus as a RAID controller, has been known to fix these errors. If these errors occur, please contact your hardware manufacturer.

http://support.veritas.com/docs/255501

[Hopper]

This is not a hardware problem. This drive has been used on the same server with 3 prior versions of BE.  The new BE V12.5 drivers for this library have just now started reporting this error.

[Anup]

Check Hardware compatiblity list

http://support.veritas.com/docs/306831

Check the version of driver you are using.

Update the driver.

also see if it works with OEM drivers.

[Hopper]

The device is a HP DAT72x6 tape library that is fully supported by BE V12.5. I am using the latest HP drivers for the SCSI card and Symantec drivers for the tape library and drive.  The SCSI adapter, the library, and the tape drive itself all have the latest HP firmware.

I haven't wanted so say this so blantently before, but I believe this is 100% a Symtantec driver problem.  We have always have had better luck with Symtantec library and tape drivers in the past, but maybe I should take your advise and use HP or MS drivers - but this also brings in other vendors for support issues like I am having now.

And what is worse, is I have no errors at all running the backups - they work just fine.  It appears to be an impatient driver programmer that did not use the recommended HP device timeouts for library operations like moving a tape from a slot to a drive.

[Larry_Fine]

Does the inventory job complete successfully?  If not, what error does the job log report?

[Hopper]

To answer Larry's question: the inventory job completed without any errors being reported by BackupExec.  Only the 6 timeout events in the windows event logs.

I tried some experimentaton.  I installed the latest firmware in the DAT library avaialble from HP (I was about two revs behind). This did "not" fix or change the problem.  I then replaced the Symantec DAT library drivers with the lastest version from HP. This did "not" fix or change the problem.

I am inclined to go look for Adaptec SCSI controller firmware/driver updates, but I do not beleive this will fix the problem - especially after the updates/changes I have already made.

I beleive the problem is still a coding issue with the way BE V12.5 impatiently handles the I/O commands to process the inventory operation.

[Larry_Fine]

I talked to a friend using that changer with BEWS 12.5 and they are not seeing the issue.