[
https://issues.apache.org/jira/browse/HBASE-10909?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13974272#comment-13974272
]
Ted Yu commented on HBASE-10909:
--------------------------------
On page 4, the box below HM1 and HM2 seems to embody coordination engine.
Where would coordination engine reside ? Does each master run an instance of
coordination engine ?
bq. RS2 will start opening R1 without worrying about possibility for it to be
hijacked in the middle of operation
What if RS2 goes into GC pause or experiences network partition ? How would the
cluster needs to respond to such situation ?
Page 5, bullet:
bq. Multiple consistent replicas of the same region may be hosted by different
region servers
HBASE-10070 proposes a solution how the above should be handled.
bq. WAL splitting may run in the background without regions downtime.
In that case, how would read requests to the underlying regions be handled ?
Page 7:
bq. abstract handlers on HM side (OpenRegionHandler
Typo: OpenRegionHandler -> OpenedRegionHandler
> Abstract out ZooKeeper usage in HBase
> -------------------------------------
>
> Key: HBASE-10909
> URL: https://issues.apache.org/jira/browse/HBASE-10909
> Project: HBase
> Issue Type: Umbrella
> Components: Zookeeper
> Reporter: Mikhail Antonov
> Assignee: Mikhail Antonov
> Attachments: HBaseConsensus.pdf, HBaseConsensus.pdf,
> HBaseConsensus.pdf
>
>
> (Original discussion started in comments for HBASE-10866)
> As some sort of follow-up or initial step towards HBASE-10296.
> Whatever consensus algorithm/library may be the chosen, perhaps one of first
> practical steps towards this goal would be to better abstract ZK-related API
> and details, which are now throughout the codebase (mostly leaked throuth
> ZkUtil, ZooKeeperWatcher and listeners).
> This jira is umbrella for relevant subtasks.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
