[
https://issues.apache.org/jira/browse/FLINK-14052?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17283677#comment-17283677
]
Till Rohrmann commented on FLINK-14052:
---------------------------------------
In general I think that this would still be nice to clean up. However, given
that we didn't do it for 1,5 years, I would say that we close it for the time
being.
> Generalize AkkaOptions to RpcOptions
> ------------------------------------
>
> Key: FLINK-14052
> URL: https://issues.apache.org/jira/browse/FLINK-14052
> Project: Flink
> Issue Type: Improvement
> Components: Runtime / Configuration, Runtime / Coordination
> Affects Versions: 1.9.0, 1.10.0
> Reporter: Till Rohrmann
> Priority: Major
>
> At the moment Flink's rpc configuration options are based on Akka specific
> configuration options defined in {{AkkaOptions}}. For example, the rpc
> timeout is configured via {{akka.ask.timeout}}. This has historical reasons
> because we introduced the {{RpcService}} after using Akka directly for rpc
> communication.
> I think this confusing for users since the fact that we use Akka should only
> be an implementation detail. Instead we should have general rpc configuration
> options such as {{rpc.timeout}} which apply to all {{RpcService}}
> implementations and implementation specific configuration options such as
> {{rpc.akka.transport.heartbeat.interval}}.
> Within the scope of this issue we could also rethink whether we don't copy
> the Akka specific configuration options names. For example
> {{rpc.akka.transport.heartbeat.interval}} should then be called
> {{rpc.akka.transport-failure-detector.heartbeat-interval}}. That way it would
> also be easy to simply parse and set them.
> As a first step we need to list all RPC relevant configuration options
> currently used and decide whether it a RPC general option or an Akka specific
> one.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
