[
https://issues.apache.org/jira/browse/FLINK-7106?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Till Rohrmann updated FLINK-7106:
---------------------------------
Component/s: (was: flink-contrib)
(was: Local Runtime)
Distributed Coordination
> Make SubmittedJobGraphStore implementation configurable
> -------------------------------------------------------
>
> Key: FLINK-7106
> URL: https://issues.apache.org/jira/browse/FLINK-7106
> Project: Flink
> Issue Type: Improvement
> Components: Distributed Coordination
> Affects Versions: 1.3.1
> Reporter: Chen Qin
> Assignee: Chen Qin
> Priority: Minor
>
> Current SubmittedJobGraphStore is hardcoded to store in zookeeper if user
> choose HAMode. The goal of this task is to allow user build their own
> implementation and pass configuration from flink.conf and define how/where
> those information stores. (e.g rocksdb statebackend)
> P.S I think this would be interesting to see how flink in HA mode can
> fallback to Standalone when zk suffers temp outages.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
