metaswirl commented on pull request #18169: URL: https://github.com/apache/flink/pull/18169#issuecomment-1013909245
> Thanks for explaining how the Yarn logic works @metaswirl. I think this is a good solution and works. One last question is whether we want to add an IT test that ensures that a TM deregisters from a `JM`. I tried testing this. But, in my scenarios, the TaskManager often received the SIGTERM after the ResourceManager had already exited. I tried to artifcially delay the termination of the ResourceManager. But I must have done it in a wrong way, because the message from the TaskManager was not received by the RM. (I saw the Akka message being sent, but not received.) I added tests that check the termination completes as long as (a) no resources are simultaneously released, (b) resources are successfully simultaneously released and (c) resources are unsuccessfully simultaneously released. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
