[
https://issues.apache.org/jira/browse/FLINK-2667?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Till Rohrmann closed FLINK-2667.
--------------------------------
Resolution: Won't Do
Closed for inactivity.
> Rework configuration parameters
> -------------------------------
>
> Key: FLINK-2667
> URL: https://issues.apache.org/jira/browse/FLINK-2667
> Project: Flink
> Issue Type: Improvement
> Components: Deployment / Scripts
> Affects Versions: 0.10.0
> Reporter: Maximilian Michels
> Assignee: Maximilian Michels
> Priority: Major
>
> In the course of FLINK-2641, we came up with some changes to make the
> configuration parameters more meaningful and self-explanatory. Some concerns
> about backwards-compatibility were raised in the associated pull request:
> https://github.com/apache/flink/pull/1125. That's why I decided to delay
> those changes.
> Here are the changes to the configuration which I would like to propose:
> {{taskmanager.memory.size}} --> {{taskmanager.memory.managed}}
> {{taskmanager.memory.fraction}} --> {{taskmanager.memory.managed.fraction}}
> {{taskmanager.heap.mb}} --> {{taskmanager.memory}}
> (Change its meaning to combined JVM heap + off-heap memory)
> {{jobmanager.heap.mb}} --> {{jobmanager.memory}}
> {{taskmanager.network.numberOfBuffers}} --> {{taskmanager.network.memory}}
> (Specify the network size in terms of space and not in terms of the number of
> buffers)
> I think those changes would make configuration easier and improve the overall
> user experience of Flink. The drawback is that it requires some users to
> update their config files. I believe the negative impact will be very little
> because only two of the changed parameters are present in the default config.
> Adapting to these new parameters should be easy because they make a lot more
> sense but I wanted to put this up for debate.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
