I have a RHEL VM which has 4 x 500GB disks attached to it from the host machine - which it serves from a 3-node RAID5 storage disk. I didn't build in redundancy on the LV layer, because the disks were presented from a RAID5 disk by the host ... stupid me for ever trusting windows.
I also attached a 512GB NVM disk image from the host, which I implemented as a cachepool for the LV - Everything was working pretty well up until the weekend.
Then - windows being windows ... one of the drives develops a fault and long story short, I cannot recover the RAID5 disk fully - After some kind of partial recovery, I have 3 of the 4 VM disk images, but one of them is corrupt and virtualbox can no longer add it to the VM.
My thinking here is - The LV was only using 200GB and the cache disk reported about 400GB used, after I copied my data onto the LV, and it had been running for a few weeks... so perhaps somehow, everything might still be on the cache disk ?
Question is - when i start up the VM, the LV will not come up because there is a PV missing. Is there any way to grab what is on the cache disk and either output to a new LV or the existing LV?
I was wondering if I could create a new PV with the same UUID and add it to the pool, tricking LVM to think that the disk is there, but because the data is striped across all the disk in the VG - it's going to have major issues.
I have not started the VM yet since the loss of one of the PVs and just wondering if anyone has any thoughts on this.
Thanks
