Github user vanzin commented on the pull request:
https://github.com/apache/spark/pull/8737#issuecomment-142135453
Ok, I think I read too much into what was going on. But:
> Also all the containers (executors) related to the old AM will be exited.
I see. That, though, seems to be caused by the reset state in the AM, not
because the executors depend on the AM in any way; the AM will send a request
to YARN that basically means "I need way less executors than I currently have,
so feel free to kill all the others".
If some the state was somehow kept between AM instances (or updated once
the new AM registers with the driver), that could be avoided. But since this
really only affects yarn-client mode, that seems like not enough value for the
extra work.
Given that, LGTM aside from the minor method rename.
---
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]
