You are viewing a plain text version of this content. The canonical link for it is here.
Posted to reviews@spark.apache.org by GitBox <gi...@apache.org> on 2019/03/21 07:19:37 UTC

[GitHub] [spark] liutang123 commented on issue #24131: [SPARK-27192][Core] spark.task.cpus should be less or equal than spark.executor.cores

liutang123 commented on issue #24131: [SPARK-27192][Core] spark.task.cpus should be less or equal than spark.executor.cores
URL: https://github.com/apache/spark/pull/24131#issuecomment-475132217
 
 
   @jiangxb1987 Thanks for review.
   Sorry I didn't noticed the checking logic in `SparkConf`, but I think the checking logic is incomplete for local mode.
   For example:
   case 1:
   ```
   $SPARK_HOME/bin/spark-shell --master local[3] --conf spark.task.cpus=2 --conf spark.executor.cores=1
   ```
   local[3] decides executor's core num is 3, but in #23290's logic, exception will be thrown.
   case 2:
   ```
   $SPARK_HOME/bin/spark-shell  --master local  --conf spark.task.cpus=2 --conf spark.executor.cores=3
   scala>sc.setLogLevel("INFO")
   scala>sc.parallelize(1 to 9).collect
   ```
   You can see spark will hang after log `INFO TaskSchedulerImpl: Adding task set 0.0 with 1 tasks.` but the checking logic in #23290 can not identify this case.
   So, I think we can check the spark.task.cpus before creating TaskScheduler.

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


With regards,
Apache Git Services

---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org