[SPARK-25342] Support rolling back a result stage

• Type: Improvement
• Status: In Progress
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 3.1.0
• Component/s: Spark Core
• Assignee: Unassigned
• Reporter: Wenchen Fan
• Created: 05/Sep/18 17:30
• Updated: 01/Aug/22 12:13

Description

This is a follow up of https://issues.apache.org/jira/browse/SPARK-23243

To completely fix that problem, Spark needs to be able to rollback a result stage and rerun all the result tasks.

However, the result stage may do file committing, which does not support re-commit a task currently. We should either support to rollback a committed task, or abort the entire committing and do it again.

Issue Links

• relates to: SPARK-23243 Shuffle+Repartition on an RDD could lead to incorrect answers
• relates to: SPARK-25341 Support rolling back a shuffle map stage and re-generate the shuffle files
• links to: [Github] Pull Request #37359 (caican00)