[
https://issues.apache.org/jira/browse/FLINK-1754?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14371050#comment-14371050
]
Fabian Hueske commented on FLINK-1754:
--------------------------------------
This deadlock bug is certainly critical and it would be very good to fix it in
0.8 as well.
We don't need to fix it in the runtime (as in 0.9) but could fix it in the
optimizer and only choose sort-merge strategies if the join closes a branch.
> Deadlock in job execution
> -------------------------
>
> Key: FLINK-1754
> URL: https://issues.apache.org/jira/browse/FLINK-1754
> Project: Flink
> Issue Type: Bug
> Affects Versions: 0.8.1
> Reporter: Sebastian Kruse
>
> I have encountered a reproducible deadlock in the execution of one of my
> jobs. The part of the plan, where this happens, is the following:
> {code:java}
> /** Performs the reduction via creating transitive INDs and removing them
> from the original IND set. */
> private DataSet<Tuple2<Integer, int[]>>
> calculateTransitiveReduction1(DataSet<Tuple2<Integer, int[]>>
> inclusionDependencies) {
> // Concatenate INDs (only one hop).
> DataSet<Tuple2<Integer, int[]>> transitiveInds = inclusionDependencies
> .flatMap(new SplitInds())
> .joinWithTiny(inclusionDependencies)
> .where(1).equalTo(0)
> .with(new ConcatenateInds());
> // Remove the concatenated INDs to come up with a transitive
> reduction of the INDs.
> return inclusionDependencies
> .coGroup(transitiveInds)
> .where(0).equalTo(0)
> .with(new RemoveTransitiveInds());
> }
> {code}
> Seemingly, the flatmap operator waits infinitely for a free buffer to write
> on.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
