Description
The BulkTableUpdateCommand updates data in a Blur table by creating a snapshot of the data and reading from the files in HDFS. Then in the reduce phase of the MR job the Rows that are not updated are dropped and the Row that have changes are re-indexed. If there are two jobs occurring at the same time and the same Rows are being updated data could be lost. So the issue is to create a lock (likely in ZK) per table for the bulk table updates.