Github user jinxing64 commented on the issue:
https://github.com/apache/spark/pull/16620
@kayousterhout
Thanks a lot for the clear explanation. It makes great sense to me and help
me understand the logic a lot. Also I think the way of testing is very good and
make the code very clear. I've already refined this pr, please take a look when
tests pass.
Also with understanding of your explanation above in
>Scenario A (performance optimization, as discussed here already): This
happens if a ShuffleMapStage gets re-run (e.g., because the first time it ran,
it encountered a fetch failure, so the previous stage needed to be re-run to
generate the missing output). ...
I made #16901 to add a test that success of old attempt should be taken as
valid and corresponding pending partition should be removed. Please give a look.
---
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]
