[
https://issues.apache.org/jira/browse/FLINK-4910?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15611885#comment-15611885
]
ASF GitHub Bot commented on FLINK-4910:
---------------------------------------
Github user StephanEwen commented on a diff in the pull request:
https://github.com/apache/flink/pull/2691#discussion_r85329878
--- Diff:
flink-runtime/src/main/java/org/apache/flink/runtime/taskmanager/Task.java ---
@@ -446,6 +447,8 @@ public void startTaskThread() {
@Override
public void run() {
+ FileSystem.createFileSystemCloseableRegistryForTask();
--- End diff --
Minor: Can you move that into the beginning of the `try` block? It may be
that the task exits early (immediate cancellation) and then we would not have
to create this at all.
> Introduce safety net for closing file system streams
> ----------------------------------------------------
>
> Key: FLINK-4910
> URL: https://issues.apache.org/jira/browse/FLINK-4910
> Project: Flink
> Issue Type: Improvement
> Reporter: Stefan Richter
> Assignee: Stefan Richter
>
> Streams that are opened through {{FileSystem}} must be closed at the end of
> their life cycle. However, we found hints that some code forgets to close
> such streams.
> We should introduce i) a mechanism that closes leaking unclosed streams after
> usage and ii) provides logging that helps us to track down and fi the sources
> of such leaks.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)