[
https://issues.apache.org/jira/browse/FLINK-4954?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15615221#comment-15615221
]
ASF GitHub Bot commented on FLINK-4954:
---------------------------------------
GitHub user tillrohrmann opened a pull request:
https://github.com/apache/flink/pull/2719
[FLINK-4954] [rpc] Discard messages when AkkaRpcActor is in state
Processing.STOP
When the AkkaRpcActor receives a message while being in state
Processing.STOP it will discard
it and send an AkkaRpcException back to the caller. This replaces the old
stashing behaviour
which had the problem that it was just a best effort approach to keep all
received messages.
Distributed components should not rely on this behaviour. That's why it was
replaced with discarding messages.
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/tillrohrmann/flink removeStashing
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/flink/pull/2719.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #2719
----
commit ed72dc67e75ae20add84c54acbacf47ce5c1b215
Author: Till Rohrmann <[email protected]>
Date: 2016-10-28T12:07:28Z
[FLINK-4954] [rpc] Discard messages when AkkaRpcActor is in state
Processing.STOP
When the AkkaRpcActor receives a message while being in state
Processing.STOP it will discard
it and send an AkkaRpcException back to the caller. This replaces the old
stashing behaviour
which had the problem that it was just a best effort approach to keep all
received messages.
Distributed components should not rely on this behaviour. That's why it was
replaced with discarding
messages.
----
> Remove stashing from AkkaRpcActor
> ---------------------------------
>
> Key: FLINK-4954
> URL: https://issues.apache.org/jira/browse/FLINK-4954
> Project: Flink
> Issue Type: Sub-task
> Components: Distributed Coordination
> Reporter: Till Rohrmann
> Assignee: Till Rohrmann
> Priority: Minor
>
> Currently, the {{AkkaRpcActor}} stashes all messages when in state
> {{Processing.STOP}}. Since the stashing storage is limited this can only work
> up to a certain amount of messages and is thus only a best effort approach.
> Since the implementation of the distributed components should not rely on
> stashing messages so that they are replayed, I propose to remove this feature.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
