Wellington Chevreuil created HBASE-24302:
--------------------------------------------
Summary: Add an "ignoreTimestamps" option (defaulted to false) to
HashTable/SyncTable tool
Key: HBASE-24302
URL: https://issues.apache.org/jira/browse/HBASE-24302
Project: HBase
Issue Type: Improvement
Affects Versions: 3.0.0-alpha-1, 2.3.0, 2.2.5
Reporter: Wellington Chevreuil
Assignee: Wellington Chevreuil
Currently, when hashing and comparing values between a source and a target
table, HashTable/SyncTable always consider cell timestamp values. However, cell
timestamp values are not always relevant for client applications, so these use
cases could benefit of a more flexible comparison logic where timestamps could
be ignored.
For such scenarios, HashTable/SyncTable could have better performance, since
cells with only timestamps diverging would not be copied.
Another case that would benefit from this option is when bulk deletes are
wrongly applied at target. At the moment, HashTable/SyncTable on it's own is
not capable of syncing back the clusters, as the source Puts would have an
older TS than the delete markers in the target. That would require target to
complete major compaction on the whole table before HashTable/SyncTable could
be run.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
