[
https://issues.apache.org/jira/browse/HBASE-11568?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14308341#comment-14308341
]
Enis Soztutar commented on HBASE-11568:
---------------------------------------
+1. I checked the patch against the master patch and current code.
Two nits we can address on commit.
- RegionReplicaReplicationEndpoint.java javadoc refers to now-gone
HLogSplitter. This coming from the old patch, we can replace that with
WALSplitter. Also the InterfaceAudience should be the one in the hbase package,
not hadoop package.
- in LogReplayOutputSink.flush() method, dataAvailable.notifyAll() is wrapped
with the sync block. It is not needed since all callers for the flush() method
already has the monitor, but we can still have that as an extra guarantee,
since it is not documented well. master code already has this (I think added
after the patch), so we can keep that sync block in the branch-1 as well.
{code}
synchronized(dataAvailable) {
- dataAvailable.notifyAll();
- }
+ controller.dataAvailable.notifyAll();
{code}
> Async WAL replication for region replicas
> -----------------------------------------
>
> Key: HBASE-11568
> URL: https://issues.apache.org/jira/browse/HBASE-11568
> Project: HBase
> Issue Type: Sub-task
> Reporter: Enis Soztutar
> Assignee: Enis Soztutar
> Fix For: 2.0.0, 1.1.0
>
> Attachments: 11568-2-branch-1.txt, 11568-branch-1.txt,
> hbase-11568_v2.patch, hbase-11568_v3.patch
>
>
> As mentioned in parent issue, and design docs for phase-1 (HBASE-10070) and
> Phase-2 (HBASE-11183), implement asynchronous WAL replication from the WAL
> files of the primary region to the secondary region replicas.
> The WAL replication will build upon the pluggable replication framework
> introduced in HBASE-11367, and the distributed WAL replay.
> Upon having some experience with the patch, we changed the design so that
> there is only one replication queue for doing the async wal replication to
> secondary replicas rather than having a queue per region replica. This is due
> to the fact that, we do not want to tail the logs of every region server for
> a single region replica.
> Handling of flushes/compactions and memstore accounting will be handled in
> other subtasks.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
