Actually, I just figured out what caused this. It seems the mount point for the brick I had on the rebooted node did not mount before gluster mounted so on the other node it guessed that it should delete everything on the volume… yikes!
So, I'm curious if this is a recoverable scenario and also how I can prevent this in the future. I admit I didn't have the brick mount in the fstab but even so can we rely on the order the mount points are written in fstab?. -- Chris LeBlanc On Friday, June 8, 2012 at 12:25 AM, Chris LeBlanc wrote: > I have a replicated gluster using nfs and one of the nodes I purposely > rebooted and may have had trouble coming up. I don't know the exact details > but I am curious to know details of recovering after a volume fails and just > ends up with a directory called .landfill (which I presume is some mechanism > to prevent wacko things from happening to your data). > > For now I copied what was in .landfill locally on one of the nodes so it's > just limping along. > > What should I do? Should I be concern about the data integrity of what's in > .landslide? > > -- > Chris LeBlanc >
