[
https://issues.apache.org/jira/browse/FLINK-2945?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14986830#comment-14986830
]
Stephan Ewen commented on FLINK-2945:
-------------------------------------
This means probably that there are still open file streams when we aim to
delete the directory. Since windows blocks the files from removal from the
directory, the shutdown fails. We would need to make sure all streams are
closed before the directory is removed.
> Shutting down a local Flink instance on Windows doesn't clean up blob cache
> ---------------------------------------------------------------------------
>
> Key: FLINK-2945
> URL: https://issues.apache.org/jira/browse/FLINK-2945
> Project: Flink
> Issue Type: Bug
> Components: BlobManager
> Affects Versions: 0.10
> Environment: Windows 10, Java 7
> Reporter: Fabian Hueske
> Priority: Minor
>
> After shutting down a local Flink instance that was started with the
> {{./bin/start-local.bat}} script and on which a job was executed as suggested
> via {{CTRL-C}}, the JobManager log file shows an ERROR message:
> {code}
> 14:24:18,547 ERROR org.apache.flink.runtime.blob.BlobCache
> - BLOB cache failed to properly clean up its storage directory.
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
