Github user CodingCat commented on the pull request:
https://github.com/apache/spark/pull/2524#issuecomment-57299654
OK, Jenkins said OK
Finished the modification,
1. Removed the option for the user to choose whether the accumulator
accepts duplication (this may need a highlight in release note)
2. still de-duplicate in DAGScheduler, since we agreed to de-duplicate
Shuffle Stage and a new TaskSetManager is created for re-submitted stage, so
DAGScheduler becomes the only place we can work on
3. Still need to add a data structure in DAGScheduler, the reason is that
ShuffleMapTask does not have something like outputid which can be used to
associate a re-submit and original task....StageId + partitionId is the only
thing we can use...
---
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]
