flink-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "ASF GitHub Bot (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (FLINK-8234) Cache JobExecutionResult from finished JobManagerRunners
Date Wed, 20 Dec 2017 08:36:00 GMT

    [ https://issues.apache.org/jira/browse/FLINK-8234?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16298081#comment-16298081
] 

ASF GitHub Bot commented on FLINK-8234:
---------------------------------------

Github user GJL commented on a diff in the pull request:

    https://github.com/apache/flink/pull/5184#discussion_r157962431
  
    --- Diff: flink-runtime/src/main/java/org/apache/flink/runtime/minicluster/MiniClusterJobDispatcher.java
---
    @@ -458,7 +465,14 @@ public JobExecutionResult getResult() throws JobExecutionException,
InterruptedE
     				}
     			}
     			else if (result != null) {
    -				return result;
    +				try {
    +					return new SerializedJobExecutionResult(
    +						jobId,
    +						result.getNetRuntime(),
    +						result.getAccumulatorResults()).toJobExecutionResult(ClassLoader.getSystemClassLoader());
    --- End diff --
    
    Because the exception is serialized in `OnCompletionActions#jobFailed(JobExecutionResult);`,
I have to deserialize it here again. I wonder if this is sane?



> Cache JobExecutionResult from finished JobManagerRunners
> --------------------------------------------------------
>
>                 Key: FLINK-8234
>                 URL: https://issues.apache.org/jira/browse/FLINK-8234
>             Project: Flink
>          Issue Type: Sub-task
>          Components: Distributed Coordination
>    Affects Versions: 1.5.0
>            Reporter: Till Rohrmann
>            Assignee: Gary Yao
>            Priority: Blocker
>              Labels: flip-6
>             Fix For: 1.5.0
>
>
> In order to serve the {{JobExecutionResults}} we have to cache them in the {{Dispatcher}}
after the {{JobManagerRunner}} has finished. The cache should have a configurable size and
should periodically clean up stale entries in order to avoid memory leaks.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

Mime
View raw message