How to prevent VM from shutting down due to corruption
I have a vmdk on a disk which has failed. I've recovered as much data as possible with clonezilla (is there a better tool for this?) and have recovered the VM on a new hard drive from backups and it's running fine, but I still need to recover the new files created since last working backup (as per murphy's law, the weekly backups have failed, so I only had access to the full backup from about a month ago).
Long story short, I can mount the recovered vmdk and see the files I need. Unfortunately they are full of landmines, whereas if process touches certain drive data, VM immediately shuts down with an error:
The redo log of '/vmfs/volumes/xxx-xxx-xxx/vm/vm-00000#.vmdk' is corrupted. If the problem persists, discard the redo log.
I've tried moving up the vmdk chain and testing images as I go. Seems the corruption is present on the first (top) snapshot, not sure if others have their own corruption, or inherit the corruption down the line.
Frustrating thing is that I'm fine with certain files being corrupted / missing. We can recreate those, but there are still many good files which have to be fished for because system insists on complete shutdown when ever corruption is detected.
Is there a setting I can set to prevent default behaviour for the corrupted vmdk files? I would like the system to keep going, even if random / unreliable data is presented to the VM in lieu of uncorrupted data. Alternatively, is there a script / process, which would go through vmdk file and mark every corrupted section as valid? Again, the root OS was recovered and running, I don't care if some PDFs, etc fail to open after getting them off of the failed VMDK.