Details
-
Improvement
-
Status: Resolved
-
Minor
-
Resolution: Fixed
-
None
-
Reviewed
Description
If we have replication setup between clusters A <-> B, during production incidents where HFiles get corrupted (e.g. via hdfs missing blocks) on cluster A, we can use CopyTable job to copy the data from B
In those cases considering the data is being copied from B to A via CopyTable, it doesn't make sense to write to WALs on A using TableOutputFormat (since the mutations being copied already exist on B).
We can add a config to customise durability (https://github.com/apache/hbase/blob/master/hbase-client/src/main/java/org/apache/hadoop/hbase/client/Durability.java) within TableOutputFormat so that an operator can use it to skip WAL writes