Bilwa S T created YARN-10449:
--------------------------------
Summary: Flexing doesn't consider containers which were stopped
Key: YARN-10449
URL: https://issues.apache.org/jira/browse/YARN-10449
Project: Hadoop YARN
Issue Type: Bug
Reporter: Bilwa S T
we have use case where if worker is idle for some period of time then user
would want to shutdown the worker to release resource and request more workers
when load is more.
In case of ON_FAILURE retry policy if user gracefully shutdown worker its exit
status will be 0 so container wont be relaunched. In this case
if user try to flex up then it currently doesn't consider stopped containers
which is not correct.
i could think of two possible solutions:
1. Consider deducting succeeded containers from Number Of Containers and then
clear succeeded component if flex up/down is done.
2. Consider updating number of containers when stopped
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-dev-unsubscr...@hadoop.apache.org
For additional commands, e-mail: yarn-dev-h...@hadoop.apache.org
