[jira] [Commented] (STORM-289) Trident DRPC memory leak

["Jon logan (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/STORM-289?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13976304#comment-13976304 ] 

Jon logan commented on STORM-289:
---------------------------------

Can you try to re-word this (or someone else)? I don't quite understand what you're trying to say.

Without looking at the code though, I have not seen this behavior, nor heard of anyone having issues with this.

> Trident DRPC memory leak
> ------------------------
>
>                 Key: STORM-289
>                 URL: https://issues.apache.org/jira/browse/STORM-289
>             Project: Apache Storm (Incubating)
>          Issue Type: Bug
>    Affects Versions: 0.9.2-incubating
>            Reporter: Gangrui Li
>
> storm.trident.spout.RichSpoutBatchTriggerer.java
> Missing code "_msgIdToBatchId.put(r, batchIdVal);" in emit method.
> As a consequence, ack/fail will get null batchId for each DRPC invoking, which makes _finishConditions map larger and larger until out of memory.



--
This message was sent by Atlassian JIRA
(v6.2#6252)