GitHub user xuanyuanking opened a pull request:
https://github.com/apache/spark/pull/20930
[SPARK-23811][Core] Same tasks' FetchFailed event comes before Success will
cause child stage never succeed
## What changes were proposed in this pull request?
This is a bug caused by abnormal scenario describe below:
ShuffleMapTask 1.0 running, this task will fetch data from ExecutorA
ExecutorA Lost, trigger `mapOutputTracker.removeOutputsOnExecutor(execId)`
, shuffleStatus changed.
Speculative ShuffleMapTask 1.1 start, got a FetchFailed immediately.
ShuffleMapTask 1 is the last task of its stage, so this stage will never
succeed because of there's no missing task DAGScheduler can get.
I apply the detailed screenshots in jira comments.
## How was this patch tested?
Add a new UT in `TaskSetManagerSuite`
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/xuanyuanking/spark SPARK-23811
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/spark/pull/20930.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #20930
----
commit 2907075b43eac26c7efbe4aca5f2c037bb5934c2
Author: Yuanjian Li <xyliyuanjian@...>
Date: 2018-03-29T04:50:16Z
[SPARK-23811][Core] Same tasks' FetchFailed event comes before Success will
cause child stage never succeed
----
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
