[
https://issues.apache.org/jira/browse/HBASE-15230?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15136952#comment-15136952
]
Ashish Singhi commented on HBASE-15230:
---------------------------------------
As per the current design the replication of WAL data starts from the beginning
of the current WAL file. So in this case the table 't1' entries already present
in the current WAL will get replicated to peer 'p2'.
Didn't find any way to set the current WAL writer offset while adding the peer
in RS.
One work around we think is to roll the WAL while adding the peer in each RS.
Any other suggestions ?
> ReplicationSource is replicating existing WAL data in a newly added peer
> ------------------------------------------------------------------------
>
> Key: HBASE-15230
> URL: https://issues.apache.org/jira/browse/HBASE-15230
> Project: HBase
> Issue Type: Bug
> Reporter: Ashish Singhi
> Assignee: Ashish Singhi
>
> Scenario:
> 1. Add a peer 'p1' and enable table replication for a table 't1'
> 2. Put some data in the table 't1' and its gets replicated to peer 'p1' as
> expected.
> 3. Remove peer 'p1' and truncate the table 't1' in both source and peer
> cluster.
> 4. Now add peer 'p2' , there is no data in source cluster in table 't1' but
> in peer cluster 'p2' where table 't1' already exists, existing WAL data of
> table 't1' is getting replicated in 'p2'.
> Expectation: Table 't1' in peer cluster 'p2' should only have data which is
> inserted in source cluster table 't1' after adding peer 'p2'
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
