AnsweredAssumed Answered

Disk recovery on OS failure

Question asked by jacques on Jul 15, 2011
Latest reply on Jul 20, 2011 by Ted Dunning
With HDFS, I know that if I lose the system disk, I can rebuild the os and simply re mount the data disks and recover the blocks I have saved on that machine without additional effort.  In the case of MapR, is the same true? 

When I add a new disk, I see that MapR formats the raw device as a MapRFS drive.  If the drive is already a MapRFS drive (from a previous install), what happens? 

So, if I lost the os and then reinstalled, could I bring that node back up without having to re-replicate everything? 

Extending further: If I lost all my OS disks for all my nodes, could I just do a brand new os install on all nodes, add the existing disks, configure mapr to point at the designated cldb nodes, and then add the disks back and bring up the cluster?  In essence, I'm trying to understand what configuration from OS disks should be backed up for this type of disaster recovery.

Outcomes