[SPARK-24415] Stage page aggregated executor metrics wrong when failures

• Type: Bug
• Status: Resolved
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: 2.3.0
• Fix Version/s: 2.3.2, 2.4.0
• Component/s: Web UI
• Assignee: Ankur Gupta
• Reporter: Thomas Graves
• Created: 29/May/18 19:15
• Updated: 27/Dec/18 11:15
• Resolved: 05/Sep/18 16:52

Description

Running with spark 2.3 on yarn and having task failures and blacklisting, the aggregated metrics by executor are not correct.  In my example it should have 2 failed tasks but it only shows one.    Note I tested with master branch to verify its not fixed.

I will attach screen shot.

To reproduce:

$SPARK_HOME/bin/spark-shell --master yarn --deploy-mode client --executor-memory=2G --num-executors=1 --conf "spark.blacklist.enabled=true" --conf "spark.blacklist.stage.maxFailedTasksPerExecutor=1" --conf "spark.blacklist.stage.maxFailedExecutorsPerNode=1"  --conf "spark.blacklist.application.maxFailedTasksPerExecutor=2" --conf "spark.blacklist.killBlacklistedExecutors=true"

import org.apache.spark.SparkEnv 

sc.parallelize(1 to 10000, 10).map { x => if (SparkEnv.get.executorId.toInt >= 1 && SparkEnv.get.executorId.toInt <= 4) throw new RuntimeException("Bad executor") else (x % 3, x) }.reduceByKey((a, b) => a + b).collect()

Issue Links

• contains: SPARK-25284 Spark UI: make sure skipped stages are updated onJobEnd
• is duplicated by: SPARK-24539 HistoryServer does not display metrics from tasks that complete after stage failure
• is duplicated by: SPARK-25910 accumulator updates from previous stage attempt should not fail
• links to: [Github] Pull Request #22209 (ankuriitg)