Github user vanzin commented on the issue:
https://github.com/apache/spark/pull/18978
Regardless of whether it's possible to re-use code for other cluster
managers, I think Saisai's initial question makes sense: should this be a
client-to-AM channel or a client-to-driver channel?
I see a client-to-driver connection as more useful; it also avoids opening
a listen socket on the AM. Anything that the driver needs the AM's help with,
the YARN-specific implementation can send a message to the currently registered
AM.
At the end, you can achieve any desired goal with either approach, but
making it client-to-driver from the start might make it simpler to handle both
client vs. cluster mode (I think), and also might make it simpler to have
similar functionality for other cluster managers later.
---
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]
