GRT backup multiple VMDKs generating error \zzzVirApp000zzz

[nicolias]

Hello,

I am going through a proof of concept with BE 2010 R3 rev 5204 (all updates as of 8/11/11).  When I have GRT enabled and backup a esx based (esxi 4.1) virtual machine (win2k8R2) with multiple VMDKs and SQL (2k8 r2) installed on several of those VMDKs via mountpoints it fails.   With the standard error, "Cannot backup directory \zzzVirtApp000zzz and its subdirectories. "

A captured debug gives the following output.  

BEREMOTE: [08/10/11 16:44:33] [4996]     [fsys\brfs]          - brUtil::OpenActiveObject Error: 0xE00084AF Could not find the object master.mdf Q:\Cat13\MSSQL10_50.CAT13\MSSQL\DATA\master.mdf

BEREMOTE: [08/10/11 16:44:33] [4996]     [fsys\ohfs]          - OFF_HOST_FS::ReadObj( ) pBrUtil->ReadObject( ) Error status CODE (E00084AF)

BEREMOTE: [08/10/11 16:44:33] [4996]     [mounter]            - UpdateSqlPdi: failed to read object info 0xe000fedf

BEREMOTE: [08/10/11 16:44:33] [4996]     [fsys\shared]        - Detach from SERVERNAME

BEREMOTE: [08/10/11 16:44:33] [4996]     [fsys\shared]        - Detach from C:

BEREMOTE: [08/10/11 16:44:33] [4996]     [mounter]            - CommonMounter::MountSomeApp: UpdateSqlPdi processing error -536805665 0xe000fedf

BEREMOTE: [08/10/11 16:44:33] [4996]     | UNEXPECTED ERROR(0xe000fedf): Failed to mount Virt-App

BEREMOTE: [08/10/11 16:44:33] [4996]     | Method:IMG_PDI_FSYS::ImgPdiFsys::ChangeDir

BEREMOTE: [08/10/11 16:44:33] [4996]     | Line  :e:\nicobar\5204r.hf10\engine\fsys\ntfs\img_pdi\staticlib\..\img_pdi_chgdir.cpp (337)

BEREMOTE: [08/10/11 16:44:40] [4996]     | Dump successfully created: 'C:\Program Files\Symantec\BackupExec\BEDBG\beremote.exe_4020_6.dmp'.

The path in question Q:\Cat13\MSSQL10_50.CAT13\MSSQL\DATA\ is a mount point going to a VMDK which exists on the same VMFS volume where the primary disk and VM files exist.   There are no extends, the VMDK is a windows NTFS basic disk.  

I've read through all the forums searching for a similar issue but can't find any.  I am assuming that there is just a problem with the GRT and mount points.

Any help and advice is greatly appreciated.

[Ben_L_]

Is the VMDK for the Q drive backed up with the virtual machine?

[nicolias]

Yes, all VMDKs of the virtual machine are selected to be backed up, including the Q drive and the DATA mountpoint.  All the VMDKs are backed up succussfully.  

I am able to restore the Q:\Cat13\MSSQL10_50.CAT13\MSSQL\DATA\master.mdf file successfully as well.   The issue is strickly related to the GRT detection of files on the mountpoints. 

To futher explain the configuration.  There are 10 VMDKs for this system.   The first 6 are on scsi controller 0 and the next 3 are on scsi controller 2.  This was required to get the VM to correctly snapshot as stated in the vmware kb article 1037071.

http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=103707...

This is just a test system with basically no data, it's a brand new install of  all software (os, sql, BE).  The VMDKs are mounted in the following manor.

Hard Disk 1 -- SCSI (0:0)  mounted as C:\

Hard Disk 2 -- SCSI (0:1)  mounted as Q:\

Hard Disk 3 -- SCSI (0:2)  mounted as Q:\Cat13\MSSQL10_50.CAT13\MSSQL\DATA\

Hard Disk 4 -- SCSI (0:3)  mounted as Q:\Cat13\MSSQL10_50.CAT13\MSSQL\TempDB\ 

Hard Disk 5 -- SCSI (0:4)  mounted as Q:\Cat13\MSSQL10_50.CAT13\MSSQL\TempDB_Log\

Hard Disk 6 -- SCSI (0:5)  mounted as Q:\Cat13\MSSQL10_50.CAT13\MSSQL\UserDB\

Hard Disk 7 -- SCSI (1:0)  mounted as Q:\Cat13\MSSQL10_50.CAT13\MSSQL\UserDB_Log\

Hard Disk 8 -- SCSI (1:1)  mounted as Q:\Cat13\MSSQL10_50.CAT13\MSSQL\Files\

Hard Disk 9 -- SCSI (1:2)  mounted as Q:\Cat13\MSSQL10_50.CAT13\MSSQL\Backups\

[Ben_L_]

The mount points are what is causing the issue. I really can't go into any more detail then that at this point, but there are problems with the VDDK following points that we are trying to work around.

This technote will be updated once the issue with mount points has been resolved.

http://www.symantec.com/docs/TECH152703

[nicolias]

Thank you very much.  I am glad to have the clarification and confirmation.

[Stefan_K_]

Hi,

I get the same error with a W2008 x64 Std., SQL 2008 R2 x64 VMWare-Machine with two virtual hard disks on BE 2010 R3 (all updates).

Nothing special about the configuration - 2 virtual hard disks, one is C:, one is D:, both NTFS. SQL-Data partly resides on D:.

Maybe the only special thing about it:
We had two disks first, added one virtual hard disk.
Then we moved the data from the second to the third hard disk.
Second hard disk was deleted, drive letter of third disk was changed to second disk's drive letter.

Any ideas? (already changed scsi-nodes)

Bye and good luck,

Stefan

[ZeRoC00L]

Stefan-K,

is your added disk also an VMDK file ?
Or is it an RDM (Raw Device Mapping) ?

[Stefan_K_]

Hi,

it's a usual .vmdk.

Bye
Stefan