Github user kayousterhout commented on the issue:
https://github.com/apache/spark/pull/17166
Thinking about this more, this seems like two separate changes (that should
probably be separated):
(1) Allowing cancellations to be injected via SparkContext. This seems
like it should have its own JIRA, and is relatively few LOC (so should be easy
to decouple). Those changes look fine and I think are good to merge as-is if
you move them to a new PR.
(2) Allowing reasons to be specified. This changes the API and changes
many LOC. I'm skeptical of this change: I think this could be helpful if
descriptive reasons are allowed (like the few I suggested in the comments), but
if you restrict reasons to a few words so that they fit in the stage summary
page, they don't seem very useful to a user. E.g., the default message of
"cancelled" when sc.killTask is used seems pretty meaningless (and will require
someone to read the code to understand -- at which point it seems like they
might as well look in the logs instead of getting info from the UI). This
doesn't seem useful enough to merit an API change, but maybe I'm missing
something important here?
---
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]
