[
https://issues.apache.org/jira/browse/FLINK-8620?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16425439#comment-16425439
]
ASF GitHub Bot commented on FLINK-8620:
---------------------------------------
Github user zentol commented on a diff in the pull request:
https://github.com/apache/flink/pull/5580#discussion_r179123350
--- Diff:
flink-runtime/src/test/java/org/apache/flink/runtime/filecache/FileCacheDirectoriesTest.java
---
@@ -164,16 +165,13 @@ public void testDirectoryCleanUp() throws Exception {
assertTrue(fs.exists(cacheFile));
fileCache.releaseJob(jobID, attemptID2);
- // still should be available, file will be deleted after 5
seconds
+ // still should be available, file will be deleted after 200
milliseconds
assertTrue(fileStatus.isDir());
assertTrue(fs.exists(cacheFile));
--- End diff --
what you _could_ also do is pass in a `ScheduledExecutorService`. You could
then intercept the delete process, verify the interval arguments and fire the
process at your leisure.
I'm worried that in it's current form the test is either not stable or
takes longer than it should.
> Enable shipping custom artifacts to BlobStore and accessing them through
> DistributedCache
> -----------------------------------------------------------------------------------------
>
> Key: FLINK-8620
> URL: https://issues.apache.org/jira/browse/FLINK-8620
> Project: Flink
> Issue Type: New Feature
> Reporter: Dawid Wysakowicz
> Assignee: Dawid Wysakowicz
> Priority: Major
>
> We should be able to distribute custom files to taskmanagers. To do that we
> can store those files in BlobStore and later on access them in TaskManagers
> through DistributedCache.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
