Products & Services
MapR Book Club
to create and rate content, and to follow, bookmark, and share content with other members.
Disk rebuilds in MapRFS
Question asked by
on Jun 12, 2012
on Jun 12, 2012 by peterconrad
Show 0 Likes
How do disk rebuilds happen in MapRFS? Suppose I lose a disk in one of the data servers how is the data recovered?
No one else has this question
Mark as assumed answered
This content has been marked as final.
Show 1 comment
(Required, will not be published)
Jun 12, 2012 9:53 AM
As you know, data cannot always be recovered from a failed disk. If you have a 100-node cluster, and disk drives have a MTBF of 3 years, you'll have a disk failing every 3 days or so on average. There are indeed tools to recover data from a failed disk, but it's unwise to put all your eggs in that basket. That's the point of replication and other features--to make sure the cluster can tolerate hardware failure.
If a disk fails, the data does not have to be rebuilt--it's present elsewhere in the cluster. When you replace the disk, you re-format the entire storage pool it was part of (usually a total of three disks) and the data is re-replicated to take advantage of the re-added storage pool.
Show 0 Likes
Retrieving data ...
Is it possible to run an Oozie Spark Action without specifying inputDir & outputDir
6.0.1: Using MapR Installer after manual install
Spark Jobs in MapR Data Science Refinery Stall Forever
Consumer Group Hang - MapR Streams with all partitions
Spark on Yarn job fails when launching container