[ https://issues.apache.org/jira/browse/FLINK-9027?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16409515#comment-16409515 ]
ASF GitHub Bot commented on FLINK-9027: --------------------------------------- Github user tillrohrmann commented on the issue: https://github.com/apache/flink/pull/5740 I agree that having a single shut down hook for cleaning up the temp directory is the best solution. I think that the `TaskManager` does not properly do it like you've described. The `TaskExecutorLocalStateStoresManager` registers its own shut down hooks and most of the `TaskManager` components like `BlobCache`, `FileCache` and `IOManager` do so likewise. The upload directory is a good point. With my changes it will always be deleted. If this is not the intended behaviour, then I will adapt it accordingly. In order to not widen the scope of this task too much I would suggest the following: 1. Addressing the general problem of reducing the number of shut down hooks in a separate issue 2. Creating a a sub directory in `WebOptions#TMP_DIR` which is set up and cleaned up by the `ClusterEntrypoint` and its shut down hook What do you think @StephanEwen. > Web UI does not cleanup temporary files > --------------------------------------- > > Key: FLINK-9027 > URL: https://issues.apache.org/jira/browse/FLINK-9027 > Project: Flink > Issue Type: Bug > Components: Webfrontend > Affects Versions: 1.5.0 > Reporter: Nico Kruber > Assignee: Till Rohrmann > Priority: Blocker > Fix For: 1.5.0 > > > The web UI creates two directories in {{java.io.tmp}}, namely > {{flink-web-<uuid>}} and {{flink-web-upload-<uuid>}} and both are not cleaned > up (running {{start-cluster.sh}} and {{stop-cluster.sh}}. -- This message was sent by Atlassian JIRA (v7.6.3#76005)