Github user jerryshao commented on the pull request:
https://github.com/apache/spark/pull/8737#issuecomment-142130226
Hi @vazin, according to my test, when AM is failed and restarted by Yarn
RM, all the internal states will be refreshed (including YarnAllocator), since
it is a new process now. Also all the containers (executors) related to the old
AM will be exited. So to some extent AM side of executor management is reset to
the initial state, what we should care is to rest driver side
`ExecutorAlloationManager` back to the initial state.
So from my understanding we don't need to take care of `YarnAllocator`,
what we need to care about is driver side state.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
