Daniel Lanza García commented on SPARK-23474:

Thanks [~gsomogyi]

Does your app have all output operations async? it is the case in mine.

My job is on YARN, I do not know if this can affect.

> mapWithState + async operations = no checkpointing
> --------------------------------------------------
>                 Key: SPARK-23474
>                 URL: https://issues.apache.org/jira/browse/SPARK-23474
>             Project: Spark
>          Issue Type: Bug
>          Components: DStreams
>    Affects Versions: 2.2.1
>            Reporter: Daniel Lanza García
>            Priority: Major
> In my Spark Streaming job I use mapWithState which obligates me to enable 
> checkpointing. A job is trigger in each batch by the operation: 
> stream.foreachRDD(rdd.foreachPartition()).
> Under this situation the job was checkpoinitng every 10 minutes (batches of 1 
> minute).
> Now, I have changed the output operation to async: 
> stream.foreachRDD(rdd.foreachPartitionAsync()).
> But checkpointing is not taking place... I tried checkpointing the RDD which 
> I map with state, it get checkpointed but does not break the lineage so tasks 
> keeps growing with every batch.

This message was sent by Atlassian JIRA

To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org

Reply via email to