[ 
https://issues.apache.org/jira/browse/SPARK-18564?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15766561#comment-15766561
 ] 

Vladimir Pchelko edited comment on SPARK-18564 at 12/21/16 9:21 AM:
--------------------------------------------------------------------

Currently the user can modify interval of mapWithState checkpoint, for example:
{code}
val CUSTOM_CHECKPOINT_DURATION_MULTIPLIER = ...
val stateDStream = sourceDStream.mapWithState(...)
stateDStream.checkpoint(batchInterval * CUSTOM_CHECKPOINT_DURATION_MULTIPLIER)
{code}


was (Author: vpchelko):
Currently user can modify interval of mapWithState checkpoint, for example:
{code}
val CUSTOM_CHECKPOINT_DURATION_MULTIPLIER = ...
val stateDStream = sourceDStream.mapWithState(...)
stateDStream.checkpoint(batchInterval * CUSTOM_CHECKPOINT_DURATION_MULTIPLIER)
{code}

> mapWithState: add configuration for DEFAULT_CHECKPOINT_DURATION_MULTIPLIER
> --------------------------------------------------------------------------
>
>                 Key: SPARK-18564
>                 URL: https://issues.apache.org/jira/browse/SPARK-18564
>             Project: Spark
>          Issue Type: Improvement
>            Reporter: Daniel Haviv
>
> Currently mapWithState checkpoints the whole state every 10 batches.
> Large state checkpointing can cause huge delays. exposing 
> DEFAULT_CHECKPOINT_DURATION_MULTIPLIER as a configuration parameter can the 
> user mitigate these delays.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

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

Reply via email to