Recovery Vmfs !!install!! <FHD>

Create a new partition (assuming a standard 1MB offset):

vmkfstools -V -r /vmfs/devices/disks/naa.6000c29c4c5a2b1c:1 If the metadata is corrupt (files show up, but VMs won't power on), you need the VMFS Offline Metadata Analyzer ( voma ).

esxcfg-volume -l # Find the volume name esxcfg-volume -U [Volume_Name] Run the repair. You must unmount the volume first, or you will crash the host. recovery vmfs

partedUtil getptbl /vmfs/devices/disks/naa.6000c29c4c5a2b1c If the output is gibberish or empty, the partition table is zeroed out. You need to know the old partition size. Usually, a VMFS datastore uses the entire LUN (Partition starting at sector 2048).

voma -m vmfs -f fix -d /vmfs/devices/disks/naa.6000c29c4c5a2b1c:1 voma will scan every file descriptor, fix chain corruption, and rebuild the allocation map. This tool has saved my bacon more times than I can count. Once the repair completes (or the mount succeeds), rescan again and verify the heartbeat. Create a new partition (assuming a standard 1MB

esxcfg-scsidevs -l Look for the device with the correct size and LUN number that does not have a filesystem label next to it. Now, we attempt a manual mount. Use the -r flag for read-only to ensure we don't make the situation worse.

partedUtil setptbl /vmfs/devices/disks/naa.6000c29c4c5a2b1c gpt partedUtil add /vmfs/devices/disks/naa.6000c29c4c5a2b1c 1 2048 [End_Sector_Number] To find the end sector: partedUtil getUsableSectors /vmfs/devices/disks/naa.6000c29c4c5a2b1c partedUtil getptbl /vmfs/devices/disks/naa

First, unmount the datastore (force if necessary).