[11.5 SP4] strange warnings and duplicated library

[FedericoZanardi]

Hi, we are experiencing some problems with our backup enviroment composed by:
IBM x346 server (with 2 Adaptec 29320ALP SCSI controllers)
A TS3200 (IBM 3573-TL libray) with 2 IBM ULT3580-TD4 drives
Libray is configred to use 1 Logical drive and failover path activated.
BrightStor ARCserve 11.5 SP4
Windows 2003 SP2 x86

Here is the summary of our problems:

- Every time we run a backup, it completes correctly logging a LOT of warnings like:
W6700 02/07/2008 11.20.40 163 MUX: Session = 2 recovered from the previous error. Detected the file mark successfully. Current TAPE: name = F-SERVER-WED-02/07/08, randomID = 4ec7, sequence = 1
W6699 02/07/2008 11.19.06 163 MUX: Session = 2 could not detect a file mark just written. So shall try to detect again. Current TAPE: name = F-SERVER-WED-02/07/08, randomID = 4ec7, sequence = 1

- Tape log reports these errors:
2008/07/02 11:16:50 [05a0] Mux_Update_ClientEndSession:: While Verifying FIleMark written, pfTapeSpaceReverse returned success instead of TP_FIL. This is an error = 0x0 -- jID[164], mID[7459046], jTKN[1214990042], jNUM@29@, dcb[1], sNUM[5]
2008/07/02 11:16:50 [05a0] Mux_Update_ClientEndSession:: Let's rewind and then space to EOD. Maybe this will cause the drive to flush. -- jID[164], mID[7459046], jTKN[1214990042], jNUM@29@, dcb[1], sNUM[5]
2008/07/02 11:16:50 [05a0] ISSUE REWIND -- jID[164], mID[7459046], jTKN[1214990042], jNUM@29@, dcb[1], sNUM[5]
2008/07/02 11:17:00 [0ffc] GetMuxLock:: Timeout Occurred. --hJob[p], jID[16941040], mID[163], jTKN[7459046], jNUM@28@, dcb[1], sNUM[2]
2008/07/02 11:17:00 [0f54] GetMuxLock:: Timeout Occurred. --hJob[p], jID[16942912], mID[165], jTKN[7459046], jNUM@30@, dcb[1], sNUM[4]
...

- And finally ARCserve show a duplicated tape library.

http://img113.imageshack.us/my.php?image=cmbg20080620brightstordty5.jpg

What we have already done?

- Tape libraries and tape drives are disabled in device manager

http://img133.imageshack.us/my.php?image=cmbg20080620windevicemabh8.jpg

- BAB device configuration SEES correctly only 1 logical library, but BAB sees 2.

http://img133.imageshack.us/my.php?image=cmbg20080620deviceconficj2.jpg

- HideDuplicatedDevices key has been correctly setted

http://img113.imageshack.us/my.php?image=cmbg20080623regeditjs8.jpg

- ALL tape library , drives and controllers has been replaced by IBM support
- Software has been reinstalled


IBM says it is a software issue
CA says it is a hardware issue
...

I need help

Federico

 

[news123]

Download the OEMCert.x86_6135.zip from:
ftp://ftp.software.ibm.com/storage/devdrvr/Windows/OEM_Drivers/

Extract the zip file, and run Install.exe (You MUST run INSTALL.EXE)

Add this registry setting to all three of the drivers:
(IBM broke up their tape drive driver into three separate drivers)
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\ibmrmsst
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\ibmtp2k3
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\ibmtpft2k3

Highlight the key, select Edit Add Value,
the Value Name is DisableReserveUponOpen,
the Type is REG_DWORD Data is 1

3. Reboot the server.

 

[FedericoZanardi]

I've installed yet the lastest certified IBM drivers, but I
can find only keys:

HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\ibmtp2k3
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\ibmtpft2k3

I could try such modify, but what is the meaning of adding
these keys? I googled the key name but I did not got a clear
explanation

thanks!

 

[vschumpy]

Drivers are not required so it is pointless installing them unless you are using StorPort drivers for the HBA.

Does the OS see two drives or one?

Is persistant binding used by the HBA?

Have you tried deleting the tape engine config reg hive and re-running device configuration?

see the end of this doc for info :

http://supportconnectw.ca.com/public/storage/infodocs/basb11win/TEC342813.pdf

 

[FedericoZanardi]

OS sees two drives (correctly) and 2 libraries (which could
be considered correctly as the library is attached through 2
different scsi controllers)
It seems that ARCserve does not correctly recognize failover
path for the library.

The backup errors has gone away since we cancelled the
registry key and rerun the device configuration.
But the library is still seen twice.

what do you mean with "Is persistant binding used by the HBA?" ?

 

[vschumpy]

OS sees two drives (correctly) and 2 libraries (which could
be considered correctly as the library is attached through 2
different scsi controllers)"

Therein lies your problem, if the OS is seeing two libraries, then so will ARCserve. It's really up to your failover software to manage the visibility of the library - it should only be seen once. Solve this problem with the OS seeing two libraries, and I'm sure you'll also resolve the device configuration issue with ARCserve.