SAN LUN corrupted reverted existing Datastore to LinuxNative from VMFS and cannot mount without VMware wiping the datastore

[zoics]

All, I have a Datastore that will not mount.  It is on a HP P2000 and VMware (5.5.0 u3b) on the host required a rebuild after corruption.  Note the store would not mount previously.  9 out of 10 stores mount and are visible in VMware. 

I have already tried to reset the partition table - using partedUtil.  following the guidance  here - VMware Knowledge Base

~ # partedUtil getptbl /vmfs/devices/disks/naa.600c0ff00014c353c7c00f5001000000

gpt

497976 255 63 8000000000

1 2048 7999997951 EBD0A0A2B9E5443387C068B6B72699C7 linuxNative 0

getUsableSectors returned -  7999999966

Current status -

~ # partedUtil getptbl /vmfs/devices/disks/naa.600c0ff00014c353c7c00f5001000000

gpt

497976 255 63 8000000000

1 2048 7999997951 AA31E02A400F11DB9590000C2911D1B8 vmfs 0

Note - have tried both 7999997951 and 7999999966 for usable sectors

The drive does not show within /vmfs/volumes yet the device shows as mounted via the vSphere Client, under storage adapter, Operational State. 

Errors in the kernel log show - 'No FS driver claimed device 'naa.600c0ff00014c353c7c00f5001000000:1': Not supported'

Calculated value for usable sectors was - 7999984439 - and this also fails.

This volume is 4TB.

Any ideas?  Many thanks.

[continuum]

> 'No FS driver claimed device 'naa.600c0ff00014c353c7c00f5001000000:1':

This usually means that there is no quick and easy way to fix this.

As you are using 5.5 you may be lucky and read the LUN via Linux vmfs-tools.

For a detailed analysis I would need the result of
dd if=/vmfs/devices/disks/naa.600c0ff00014c353c7c00f5001000000 of=/vmfs/volumes/some-datastore/zoics.1536  bs=1M count=1536

Ulli

[a_p_]

Just one quick question. Did you run the command from step 1 "Find the beginning and end blocks of the VMFS partition.", and what exactly did it return for the missing datastore/device?

André

[zoics]

Ulli Many thanks for your response.  I have exported the data as requested, unfortunately after viewing the content with a Hex Editer I realise I cannot release this data outside the organisation.

I tried mounting the image with OFSmount - which identified the partition and the entire disk, and selecting either results in an inaccessible drive, with suggestion it is corrupt.

Are you able to indicate the steps you would undertake with the output of the dd command?

[zoics]

Andre - the output is below, but this was not carried out at the outset. I have included the datastore before the broken one for reference -

---------------------

/vmfs/devices/disks/naa.600c0ff00014c3538c9a065001000000

gpt

248988 255 63 4000000000

1 2048 3999999966 AA31E02A400F11DB9590000C2911D1B8 vmfs 0

Checking offset found at 2048:

0200000 d00d c001

0200004

1400000 f15e 2fab

1400004

0140001d  44 61 74 61 00 00 00 00  00 00 00 00 00 00 00 00  |Data............|

0140002d  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  |................|

---------------------

/vmfs/devices/disks/naa.600c0ff00014c353c7c00f5001000000

gpt

497976 255 63 8000000000

1 2048 7999999966 AA31E02A400F11DB9590000C2911D1B8 vmfs 0

---------------------

[continuum]

> Are you able to indicate the steps you would undertake with the output of the dd command?
1. check wether the VMFS magic bits are at the expected location
2. check wether the hidden .sf files are present
3.  try wether mount via vmfs-tools is possible
- if that is possible you can then copy out the data yourself
4. if mount via vmfs-tools is impossible I extract vmdk-descriptorfiles and vmx-files

5. check wether I can create dd scripts for the flat.vmdks and delta.vmdks
- if that works I would give you dd-scripts which will enable you to manually extract the flat- and delta vmdks.