Was testing how glusterfs behaves after an HDD crash and came to this issue where a previously used but not corrupted brick cannot be added back to a volume because glusterfs thinks it already is part of a volume.
I know how to fix it but if such brick was used in this same volume before is it possible to "recover" it? Is it better to try and reuse or clear it and add as if it was a new brick?
This is what I did:
- dd /dev/zero over a brick
- Removed whole node from glusterfs volume (Intended to remove only corrupted brick, but giving my setup I managed to only remove the whole node containing the corrupted brick)
- Unmounted brick partition
- Formatted
- Mounted fixed brick partition
- Tried to add node back into the volume, but the "good" bricks were considered part of a volume.
- Cleared bricks like this or this
- Added node and bricks
