Github user vanzin commented on the pull request:
https://github.com/apache/spark/pull/8668#issuecomment-139101880
@KaiXinXiaoLei I'm not sure I follow you. It sounds like you may be
describing a race condition somewhere, but it's not clear.
Both the heartbeat receiver and the allocation manager will kill executors
using the same API (`killExecutors`) but with different parameters. That method
is synchronized so I don't see how there could be a race.
The only thing I can potentially come up with is that if the heartbeat
receiver kills executors (and then asks for new ones to replace them), the idle
timeout for the old executors will be lost (the new executors will start with a
new idle timer). That means those "x" executors will be alive for more time
than maybe would be optimal, but I don't necessarily see that as a problem.
---
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]
