After an accident power outage, one of the datastore is missing and couldn't be recognized by ESXi anymore.
The hard drives are local drives installed on the server directly and mapped with HP Array. The hard drive is showing healthy from the array.
In the storage -> devices, I could see the problematic disk is showing a "Normal" status, but if click on the disk, I can see the partitions are having missing attributes.
In the shell, all the partitions are marked as "unknown"
[root@esxi:~] partedUtil getptbl /vmfs/devices/disks/naa.600508b1001030353232393346301100
gpt
62256 255 63 1000149680
1 2048 4095 2168614864496E6F744E656564454649 unknown 0
2 4096 2101247 0FC63DAF848347728E793D69D8477DE4 unknown 0
3 2101248 1000146943 0FC63DAF848347728E793D69D8477DE4 unknown 0
[root@esxi:~] partedUtil fixGpt /vmfs/devices/disks/naa.600508b1001030353232393346301000
[root@esxi:~] vmkfstools -x check /vmfs/devices/disks/naa.600508b1001030353232393346301100
DiskLib_Check() failed for source disk '/vmfs/devices/disks/naa.600508b1001030353232393346301100': The file specified is not a virtual disk (15).
[root@esxi:~] voma -m vmfs -f check -d /vmfs/devices/disks/naa.600508b1001030353232393346301100:3
...
ERROR: LVM Major or Minor version Mismatch, Not supported
ERROR: Failed to Initialize LVM Metadata
VOMA failed to check device : Not Supported
Total Errors Found:
...
I also found a post talking about recovering the metadata by moving the partition. But it uses fdisk and doesn't explain how to calculate the offset. So I can't take the risk to do so.
What I can do to recover the partitions?
Thanks in advance!
Do you have a second healthy datastore that is still usable ?
If you provide a VMFS header dump for
/vmfs/devices/disks/naa.600508b1001030353232393346301100:3
then I may be able to help you. See
https://communities.vmware.com/t5/VMware-vSphere-Documents/Create-a-VMFS-Header-dump-using-an-ESXi-H...
Ulli
Thank you very much@continuum I've dumped all three partitions, just in case you need them. You can retrieve the files from this link.
Please note, I named the partition by order: p1-p3. So vmfs-header-dump-p3.2000 is the 3rd partition that you're interested in.
Your dump is:
- incomplete - mbr.bin is missing
- it does not have a readme.txt explaining what you are looking for
- you do not list the commands used to create the files
- containing stuff nobody asked for (part 1 and part 2)
I tried to download vmfs-header-dump-p3.2000 4 times using two different computers and each times the download failed.
Please dont waste my and your time.
Ulli
-
Hi @continuum deeply apologize for that. I was not familiar with the commands.
However, I have re-exported the files that you need (bin, dump and readme) and uploaded it to Dropbox instead.
Hope this time works.
Interesting - this dump was taken from an ext3/4 filesystem with a timestamp from 06.05.2021.
@continuum that's very odd. It was the datastore created by the same server . And vSphere also tells me it's inaccessible.
The datastore was created a long time ago, might match your timestamp. But it can't be an ext filesystem, unless vSphere created an ext instead of vmfs at that time??
I have an idea, if I dump the whole partition and mount it on a Linux box, will it be recognized? I will need to find a large enough disk to store such a big dump first.
Can I also ask how you analyzed the dump? How to restore the partition? Is there a reference I can read?