[
https://issues.apache.org/jira/browse/HBASE-11025?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13977855#comment-13977855
]
Mikhail Antonov commented on HBASE-11025:
-----------------------------------------
bq. Should I put back a HRS constructor that internally calls the new
constructor that takes a consensus implementation Just-In-Case?
I think actually yes, that's most safe way to do that. The ctor without
consensus param would pick the one from factory, right? Do you want to do that
or I can modify the patch?
bq. Zk could hang out then after all else has gone down till it gets its
callback?
I'd think for the case of hbase-managed-ZK we should shut down it from this
callback, but that's not something we immediately need now.
If ZK is external to HBase cluster (which is probably most often the case?),
then we don't need to do anything specific on shutdown, like we don't do
anything special now?
> Infrastructure for pluggable consensus service
> ----------------------------------------------
>
> Key: HBASE-11025
> URL: https://issues.apache.org/jira/browse/HBASE-11025
> Project: HBase
> Issue Type: Sub-task
> Components: Zookeeper
> Reporter: Mikhail Antonov
> Assignee: Mikhail Antonov
> Attachments: HBASE-11025(add param in HRS ctor).patch,
> HBASE-11025(add param in HRS ctor).patch, HBASE-11025.patch
>
>
> Related to HBASE-10915.
> In this jira I will extract the changed for property changes, factory and
> consensus provider interface (+ZK implementation), as it's going to be
> required by other subtasks of HBASE-10909.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
