[
https://issues.apache.org/jira/browse/SPARK-28917?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16930616#comment-16930616
]
Imran Rashid commented on SPARK-28917:
--------------------------------------
I finally got some more info about this case -- they are not using
checkpointing, nor touching dependencies. It seems things work consistently
once they move the call to {{rdd.cache()}} before touching the RDD with
multiple threads, so it could still be that caching alone is enough to mess
this up somehow.
Just by coincidence, another entirely separate group is reporting what looks to
be a very similar bug with submitting jobs from multiple threads. Its not
exactly the same, though -- it doesn't have the orphaned stages in the logs,
but does have repeated RDD registration.
> Jobs can hang because of race of RDD.dependencies
> -------------------------------------------------
>
> Key: SPARK-28917
> URL: https://issues.apache.org/jira/browse/SPARK-28917
> Project: Spark
> Issue Type: Bug
> Components: Scheduler
> Affects Versions: 2.3.3, 2.4.3
> Reporter: Imran Rashid
> Priority: Major
>
> {{RDD.dependencies}} stores the precomputed cache value, but it is not
> thread-safe. This can lead to a race where the value gets overwritten, but
> the DAGScheduler gets stuck in an inconsistent state. In particular, this
> can happen when there is a race between the DAGScheduler event loop, and
> another thread (eg. a user thread, if there is multi-threaded job submission).
> First, a job is submitted by the user, which then computes the result Stage
> and its parents:
> https://github.com/apache/spark/blob/24655583f1cb5dae2e80bb572604fb4a9761ec07/core/src/main/scala/org/apache/spark/scheduler/DAGScheduler.scala#L983
> Which eventually makes a call to {{rdd.dependencies}}:
> https://github.com/apache/spark/blob/24655583f1cb5dae2e80bb572604fb4a9761ec07/core/src/main/scala/org/apache/spark/scheduler/DAGScheduler.scala#L519
> At the same time, the user could also touch {{rdd.dependencies}} in another
> thread, which could overwrite the stored value because of the race.
> Then the DAGScheduler checks the dependencies *again* later on in the job
> submission, via {{getMissingParentStages}}
> https://github.com/apache/spark/blob/24655583f1cb5dae2e80bb572604fb4a9761ec07/core/src/main/scala/org/apache/spark/scheduler/DAGScheduler.scala#L1025
> Because it will find new dependencies, it will create entirely different
> stages. Now the job has some orphaned stages which will never run.
> The symptoms of this are seeing disjoint sets of stages in the "Parents of
> final stage" and the "Missing parents" messages on job submission, as well as
> seeing repeated messages "Registered RDD X" for the same RDD id. eg:
> {noformat}
> [INFO] 2019-08-15 23:22:31,570 org.apache.spark.SparkContext logInfo -
> Starting job: count at XXX.scala:462
> ...
> [INFO] 2019-08-15 23:22:31,573 org.apache.spark.scheduler.DAGScheduler
> logInfo - Registering RDD 14 (repartition at XXX.scala:421)
> ...
> ...
> [INFO] 2019-08-15 23:22:31,582 org.apache.spark.scheduler.DAGScheduler
> logInfo - Got job 1 (count at XXX.scala:462) with 40 output partitions
> [INFO] 2019-08-15 23:22:31,582 org.apache.spark.scheduler.DAGScheduler
> logInfo - Final stage: ResultStage 5 (count at XXX.scala:462)
> [INFO] 2019-08-15 23:22:31,582 org.apache.spark.scheduler.DAGScheduler
> logInfo - Parents of final stage: List(ShuffleMapStage 4)
> [INFO] 2019-08-15 23:22:31,599 org.apache.spark.scheduler.DAGScheduler
> logInfo - Registering RDD 14 (repartition at XXX.scala:421)
> [INFO] 2019-08-15 23:22:31,599 org.apache.spark.scheduler.DAGScheduler
> logInfo - Missing parents: List(ShuffleMapStage 6)
> {noformat}
> Note that there is a similar issue w/ {{rdd.partitions}}. I don't see a way
> it could mess up the scheduler (seems its only used for
> {{rdd.partitions.length}}). There is also an issue that {{rdd.storageLevel}}
> is read and cached in the scheduler, but it could be modified simultaneously
> by the user in another thread. Similarly, I can't see a way it could effect
> the scheduler.
> *WORKAROUND*:
> (a) call {{rdd.dependencies}} while you know that RDD is only getting touched
> by one thread (eg. in the thread that created it, or before you submit
> multiple jobs touching that RDD from other threads). Then that value will get
> cached.
> (b) don't submit jobs from multiple threads.
--
This message was sent by Atlassian Jira
(v8.3.2#803003)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
