[
https://issues.apache.org/jira/browse/HBASE-19665?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16391303#comment-16391303
]
Zheng Hu commented on HBASE-19665:
----------------------------------
Patch.v2: Let peer storage and queue storage share a single table
hbase:replication. the rowkey can be peerId or a serverName (As we discussed
before, serverName will not conflict with peerId because of their different
format):
if rowkey is a peerId, we have 3 column families:
* cf peer, column peer:config to store peer configuration, column peer:state
to store peer state.
* cf hfile-refs, column hfile-refs:<hfile-name> to store hfile refs in its
peer.
* cf regions family, column regions:<encoded-region-name> to store the latest
pushed sequence id of region for serial replication.
if rowKey is a serverName, we have 3 column families too:
* cf rs_state, column rs_state:enabled indicate that the server has been added
as a replicator.
* cf queue, column queue:<queue-id> indicate that the queue with queue-id has
been added to the region server. Note that its value in cell is a EMPTY_BYTES,
which is meaningless.
* cf wal, column wal:<queue-id> to store a byte array which is encoded the
<filename, position> pair.
> Add table based replication peers/queues storage back
> -----------------------------------------------------
>
> Key: HBASE-19665
> URL: https://issues.apache.org/jira/browse/HBASE-19665
> Project: HBase
> Issue Type: Sub-task
> Components: Replication
> Reporter: Guanghao Zhang
> Assignee: Zheng Hu
> Priority: Major
> Attachments: HBASE-19665.v1.patch, HBASE-19665.v2.patch
>
>
> We removed them after HBASE-19397. So open a issue to track this thing.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
