Hi all,
we had a power outage that seems to have damaged our old ESXI (5.0.0) installation, the SSD (single drive, not part of a RAID) we used for it seems unreadable and hence all guest systems are shown as "unknown".
I tried to access the SSD by attaching it to a linux machine (via USB) and to use vmfs-fuse. Unfortunately it shows me:
VMFS VolInfo: invalid magic number 0x00000000
VMFS: Unable to read volume information
Trying to find partitions
Unable to open device/file "/dev/md126".
Unable to open filesystem
The magic number looks no good, what I already found out.
We have a new server in place already (esxi 6.5.0) where I, prior to trying the above, attached the drive, activated it as volume on the controller (without initialising it) and bootet it up. Esxi recognizes the drive but NOT the data on it, so no partition information / datastore. It could be due to the big version jump or just because the drive is toast.
Any chances I can recover the vmdk files on the drive? Anything left I could try? We have backups on client-level, yes, but the data of the last almost 24 hours is not part of the backup, so I would really appreciate getting at least the content of the databases from there.
Any help and/or hint are highly appreciated.
Michal asked me to investigate the case.
VMFS-Metadata is partially corrupt so mounting via vmfs-fuse was impossible.
So I analysed the VMFS--metadata and was able to create dd scripts to extract the vmdks manually.
We are now waiting for the results and so far we found no reasons to assume that the ssd itself is corrupt.
We will have the first VM back in about an hour ....
Ulli
As you are not able to access data from linux machine and from other host also, Looks like SSD is corrupted.
Michal asked me to investigate the case.
VMFS-Metadata is partially corrupt so mounting via vmfs-fuse was impossible.
So I analysed the VMFS--metadata and was able to create dd scripts to extract the vmdks manually.
We are now waiting for the results and so far we found no reasons to assume that the ssd itself is corrupt.
We will have the first VM back in about an hour ....
Ulli
UPDATE: first VM is running again - no dataloss
The first VM is running again and the other two are on their way to do so, too.
I reached out to @continuum whom I found on this board and it was the best decision I could have made. I am no noob regarding IT but what we he does is like rocket science for me. He saved me hours and hours of additional work - on a sunny saturday afternoon, one hour after I first contacted him.
5 ***** plus recommendation!
Final update:
all 3 VMs that were lost are back in production - no dataloss
Ulli