SCSI Tape Drive errors

[rmv]

I'm running Arcserve 2000 still on a w2k server with a compaq dl891dlx library. I've been having problems with my full backups that require the backup to span on several tapes. All incremental backups works fine.

In the Arcserve log:
The full backup starts
finishes one tape
grabs second tape
formats it
then I'll see scsi port errors in the log
then unable to write to media

In the Windows System event log:
Type: Error
Source: cpq32fs2
Event ID: 9
Event Time: 11/6/2005 9:16:45 PM
User: n/a
Computer: RMV1ARC02
Description:
The device, \Device\Scsi\cpq32fs21, did not respond within the timeout period.

Type: Error
Source: Storage Agents
Event ID: 1107
Event Time: 11/6/2005 11:56:36 PM
User: n/a
Computer: RMV1ARC02
Description:
Compaq SCSI Tape Drive Status Change.
The tape drive with
SCSI target 4 connected to SCSI bus 1 of the controller
in slot 2 has a new status of 5.
(Tape drive status
values: 1=other, 2=ok, 3=failed, 5=offline, 6=missingWasOk,
7=missingWasFailed, 8=missingWasOffline)
[SNMP TRAP: 5016
in CPQSCSI.MIB]


Type: Error
Source: Storage Agents
Event ID: 1107
Event Time: 11/6/2005 11:58:36 PM
User: n/a
Computer: RMV1ARC02
Description:
Compaq SCSI Tape Drive Status Change.
The tape drive with
SCSI target 4 connected to SCSI bus 1 of the controller
in slot 2 has a new status of 2.
(Tape drive status
values: 1=other, 2=ok, 3=failed, 5=offline, 6=missingWasOk,
7=missingWasFailed, 8=missingWasOffline)
[SNMP TRAP: 5016
in CPQSCSI.MIB]

I've gone through and checked termination, all hardware and there doesn't seem to be any hardware issues. I've found an article in regards to the cpq32fs2 driver and it should be running v 5.14.20.0 or later which it is. Deleted database and even recreated job. Just to mention this has been running fine for years.

Does anyone have any other suggestions?

Thanks in advance.

 

[speshalyst]

Hi,

What is the service pack level of ARCserve 2000?

Make sure you are on Sp5 with the latest device updates.

So it Shall be Written!
So it Shall be Done!!

 

[rmv]

Yes it's running SP 5 with the latest device update.

Thanks!

 

[speshalyst]

Hi,

YOu have pasted the errors on the Windows System log ...

What errors do you see on the ARCserve Activity log ?

YOu could set the tape engine in debug ...look at link for details..

http://www.tek-tips.com/faqs.cfm?fid=3964

you could also simulate a backup to span across tapes so that you dont have to create a job that "actually" spans tapes..

you could do this by setting this on the registry of the ARCserve server..

HKEY_LOCAL_MACHINE \ SOFTWARE \ ComputerAssociates \ BrightStor ARCserve Backup \ Base \ TapeEngine

Highlight "TapeEngine" then click on Edit, then select add Key.

Key Name: FAKE_EARLY_WARNING
Class: <blank>

Next highlight "FAKE_EARLY_WARNING" then click on Edit, and select add Value.

Value Name: Length
Data Type: REG_DWORD
data: c800

this would force arcserve to span to the second tape after backing up around 25 Mbs on the first.. so you could select a folder a little over 25 megs to be backed up..

see if behaviour ocurs ALWYAS when the tapes span ..

So it Shall be Written!
So it Shall be Done!!

 

[davidmichel]

make sure the tape drive and library are disabled from within the OS Device Manager. HP/Compaq Management Agents (found in Control Panel) should be configure to not collect SCSI Information.

 

[vschumpy]

If it's been running fine for years - what changed?

Also check this document:

http://supportconnect.ca.com/sc/redir.jsp?reqPage=search&searchID=TEC342813

 

[rmv]

I'll give Fake Early warning a try and yes the tape drives are disabled through device manager.

The only thing that has changed is we have changed the original job from Full backups Monday - Friday to Incremental Mon - Thurs and Full on Fridays a month back. That was the reason why I tried creating a new job thinking the job might have ben corrupted but that also was proved wrong. So I know the tape spanning has worked before.

Could the scsi driver go bad? I'll try reapplying Sp 5 and the last device update as well. It sure won't hurt at this point.

 

[davidmichel]

I have not come across a driver becoming corrupted, but the way to test that is to uninstall the driver, restart the system, and then reinstall.

Did you verify the HP Management Agent is not collecting SCSI Information?

Could also be hardware.