You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@flink.apache.org by "Stephan Ewen (JIRA)" <ji...@apache.org> on 2016/09/27 12:53:20 UTC

[jira] [Created] (FLINK-4696) Limit the number of Akka Dispatcher Threads in LocalMiniCluster

Stephan Ewen created FLINK-4696:
-----------------------------------

             Summary: Limit the number of Akka Dispatcher Threads in LocalMiniCluster
                 Key: FLINK-4696
                 URL: https://issues.apache.org/jira/browse/FLINK-4696
             Project: Flink
          Issue Type: Bug
          Components: Distributed Coordination
    Affects Versions: 1.2.0
            Reporter: Stephan Ewen
            Assignee: Stephan Ewen
             Fix For: 1.2.0


By default, akka spawns 2x or 3x the number of cores in threads.

For the LocalFlinkMiniCluster, running on Travis (often 64 cores), with separate actor systems for jobmanager and multiple taskmanagers, this frequetly means >600 akka threads. Flink uses about 4 actors.

This simply eats unnecessary resources. I suggest to have at most 4 threads per actor system in test setups.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)