[
https://issues.apache.org/jira/browse/APEXCORE-439?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15255642#comment-15255642
]
ASF GitHub Bot commented on APEXCORE-439:
-----------------------------------------
Github user amberarrow commented on a diff in the pull request:
https://github.com/apache/incubator-apex-core/pull/314#discussion_r60847135
--- Diff:
engine/src/main/java/com/datatorrent/stram/plan/physical/PhysicalPlan.java ---
@@ -979,6 +998,14 @@ public void deployChanges() {
this.undeployOpers.removeAll(newOpers.keySet());
--- End diff --
If I understand @tweise correctly, he is suggesting that the existing code
be augmented _only_ by the highlighted fragment below (since he says "remove
everything added new"):
{code}
updatePartitionsInfoForPersistOperator(this.dag);
/* begin new code */
// redeploy dependencies of the new operators excluding the new
operators themselves
this.undeployOpers.addAll(getDependents(newOpers.keySet()));
/* end new code */
this.undeployOpers.removeAll(newOpers.keySet());
//make sure all the new operators are included in deploy operator list
this.deployOpers.addAll(this.newOpers.keySet());
// include downstream dependencies of affected operators into redeploy
Set<PTOperator> deployOperators = this.getDependents(this.deployOpers);
ctx.deploy(releaseContainers, this.undeployOpers, newContainers,
deployOperators);
{code}
This works for _undeployOpers_ since the argument set (newOpers) is removed
just after the new code. However, this change alone does not add dependents to
_deployOpers_ as the proposed fix does -- not sure if this makes a difference.
> After dynamic repartition application appears blocked
> -----------------------------------------------------
>
> Key: APEXCORE-439
> URL: https://issues.apache.org/jira/browse/APEXCORE-439
> Project: Apache Apex Core
> Issue Type: Bug
> Affects Versions: 3.3.0
> Reporter: Munagala V. Ramanath
> Assignee: Munagala V. Ramanath
>
> When a file input operator (subclass of AbstractFileInputOperator) is
> dynamically repartitioned, some downstream unifiers are not identified as
> needing to be redeployed causing the window id to not advance and all
> downstream operators to be killed and restarted.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
