[
https://issues.apache.org/jira/browse/SPARK-7477?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Dibyendu Bhattacharya updated SPARK-7477:
-----------------------------------------
Description:
With Spark Streaming on Tachyon as the OFF_HEAP block store
I have used the low level Receiver based Kafka consumer
(http://spark-packages.org/package/dibbhatt/kafka-spark-consumer) for Spark
Streaming to pull from Kafka and write Blocks to Tachyon
What I see TachyonBlockManager.scala put the blocks in WriteType.TRY_CACHE
configuration . And because of this Blocks ate evicted from Tachyon Cache and
when Spark try to find the block it throws BlockNotFoundException .
When I modified the WriteType to CACHE_THROUGH , BlockDropException is gone ,
but it impact the throughput ..
was:
With Spark Streaming on Tachyon as the OFF_HEAP block store
I have used the low level Receiver based Kafka consumer
(http://spark-packages.org/package/dibbhatt/kafka-spark-consumer) to pull from
Kafka and write Blocks to Tachyon
What I see TachyonBlockManager.scala put the blocks in WriteType.TRY_CACHE
configuration . And because of this Blocks ate evicted from Tachyon Cache and
when Spark try to find the block it throws BlockNotFoundException .
When I modified the WriteType to CACHE_THROUGH , BlockDropException is gone ,
but it impact the throughput ..
> TachyonBlockManager Store Block in TRY_CACHE mode which gives
> BlockNotFoundException when blocks are evicted from cache
> -----------------------------------------------------------------------------------------------------------------------
>
> Key: SPARK-7477
> URL: https://issues.apache.org/jira/browse/SPARK-7477
> Project: Spark
> Issue Type: Bug
> Components: Block Manager
> Affects Versions: 1.4.0
> Reporter: Dibyendu Bhattacharya
>
> With Spark Streaming on Tachyon as the OFF_HEAP block store
> I have used the low level Receiver based Kafka consumer
> (http://spark-packages.org/package/dibbhatt/kafka-spark-consumer) for Spark
> Streaming to pull from Kafka and write Blocks to Tachyon
> What I see TachyonBlockManager.scala put the blocks in WriteType.TRY_CACHE
> configuration . And because of this Blocks ate evicted from Tachyon Cache and
> when Spark try to find the block it throws BlockNotFoundException .
> When I modified the WriteType to CACHE_THROUGH , BlockDropException is gone ,
> but it impact the throughput ..
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
