[
https://issues.apache.org/jira/browse/HBASE-11467?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14118746#comment-14118746
]
Mikhail Antonov commented on HBASE-11467:
-----------------------------------------
Updated the diff on RB with latest changes.
[~stack] Right, I think we need to have a dedicated discussion around master
changes/topology. On the question about
bq. If we partition the master – a far-out possibility that hopefully we never
have to do where master 1. does tables A-D and master 2. does tables E-H, etc.
– then how would this play out? The cluster string would have to list all
masters ?
I think the more robust solution would be forwarding from one master to another
(i.e., clearly each master should know full cluster topology anyway, but
clients shouldn't have to, they should be able to ask any master they know,
i.e. a client may only know about some subset of masters, and get location of
the master they actually need). More brainstorming is needed here, for sure.
> New impl of Registry interface not using ZK + new RPCs on master protocol
> -------------------------------------------------------------------------
>
> Key: HBASE-11467
> URL: https://issues.apache.org/jira/browse/HBASE-11467
> Project: HBase
> Issue Type: Sub-task
> Components: Client, Consensus, Zookeeper
> Affects Versions: 2.0.0
> Reporter: Mikhail Antonov
> Assignee: Mikhail Antonov
> Fix For: 2.0.0
>
> Attachments: HBASE-11467.patch, HBASE-11467.patch, HBASE-11467.patch
>
>
> Currently there' only one implementation of Registry interface, which is
> using ZK to get info about meta. Need to create implementation which will be
> using RPC calls to master the client is connected to.
> Review of early version of patch is here: https://reviews.apache.org/r/24296/
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
