apex-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Vlad Rozov (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (APEXCORE-769) activate() called multiple times for redeployed operator when stream is THREAD_LOCAL
Date Fri, 28 Jul 2017 14:53:00 GMT

    [ https://issues.apache.org/jira/browse/APEXCORE-769?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16105062#comment-16105062
] 

Vlad Rozov commented on APEXCORE-769:
-------------------------------------

[~francisf] Was the problem introduced in 3.6.0? Why is 3.6.0 the only version that is affected?

> activate() called multiple times for  redeployed operator when stream is THREAD_LOCAL
> -------------------------------------------------------------------------------------
>
>                 Key: APEXCORE-769
>                 URL: https://issues.apache.org/jira/browse/APEXCORE-769
>             Project: Apache Apex Core
>          Issue Type: Bug
>    Affects Versions: 3.6.0
>            Reporter: Francis Fernandes
>            Assignee: Francis Fernandes
>
> activate() is called during the redeployment of operators when the stream locality is
set to THREAD_LOCALITY. This happens for all the child operator nodes when they are redeployed
in the same container that were previously deployed. A simple way to simulate this is by killing
an upstream(non thread-local) operator's container.
> The cause for this happening is, when setupOiOGroups() is called inside StreamingContainer,
the child nodes are appended to the oioGroups values. These values were populated as a part
of the previous deployment. A possible fix is to call oioGroups.clear() at the beginning of
setOiOGroups().



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

Mime
View raw message