[ https://issues.apache.org/jira/browse/FLINK-3897?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15281352#comment-15281352 ]
Aljoscha Krettek commented on FLINK-3897: ----------------------------------------- What are the {{ForkJoinTasks}} used for. [~trohrm...@apache.org] you did some work on the {{JobManager}}, could you shed some light on this? > why does jobmanager process has too many ForkJoinTask object > ------------------------------------------------------------ > > Key: FLINK-3897 > URL: https://issues.apache.org/jira/browse/FLINK-3897 > Project: Flink > Issue Type: Improvement > Reporter: ZhengBowen > > jmap -histo 21415 |head -10 > num #instances #bytes class name > ---------------------------------------------- > 1: 18458230 5059213664 [C > 2: 29703 1947328680 [Lscala.concurrent.forkjoin.ForkJoinTask; > 3: 993331 1361835272 [B > 4: 35228671 1127317472 akka.dispatch.AbstractNodeQueue$Node > 5: 23123301 924932040 > akka.actor.LightArrayRevolverScheduler$TaskHolder > 6: 15217224 730426752 java.util.HashMap$Entry > 7: 16863612 539635584 java.lang.String > ------------------------------------------- > ForkJoinTask object takes up to 2GB of memory space,and the memory used has > been growing. > why? -- This message was sent by Atlassian JIRA (v6.3.4#6332)