Till Rohrmann created FLINK-6523:
------------------------------------
Summary: Improve lifecycle management of created BlobCaches in TaskManagerLogHandler and JobClient
Key: FLINK-6523
URL:
https://issues.apache.org/jira/browse/FLINK-6523 Project: Flink
Issue Type: Bug
Components: Distributed Coordination
Affects Versions: 1.3.0, 1.4.0
Reporter: Till Rohrmann
Some components like the {{JobClient#retrieveClassLoader}} and {{TaskManagerLogHandler}} create {{BlobCaches}} to retrieve data from the {{BlobServer}}. These instances are created in a local context and their lifecycle is not properly managed. The cleanup/closing solely relies on the fact that their storage directory will be cleaned up eventually by a registered shutdown hook.
I think it would be better to properly implement a lifecycle management for these components and close them at an appropriate location in the code. E.g. one could pass in a shared {{BlobCache}} which is closed once the JobClient action is over.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)