You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@systemml.apache.org by "Matthias Boehm (JIRA)" <ji...@apache.org> on 2017/03/10 01:35:37 UTC
[jira] [Updated] (SYSTEMML-1386) ClassCastException due to wrong
execution context w/ parfor spark and forced singlenode
[ https://issues.apache.org/jira/browse/SYSTEMML-1386?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Matthias Boehm updated SYSTEMML-1386:
-------------------------------------
Description:
Generally, we allow that users force singlenode execution (of operations) but a remote_mr or remote_spark for parfor loops. For parfor remote_spark, this currently fails with the following stracktrace due to a wrong execution context.
{code}
Caused by: java.lang.ClassCastException: org.apache.sysml.runtime.controlprogram.context.ExecutionContext cannot be cast to org.apache.sysml.runtime.controlprogram.context.SparkExecutionContext
at org.apache.sysml.runtime.controlprogram.parfor.DataPartitionerRemoteSpark.partitionMatrix(DataPartitionerRemoteSpark.java:66)
at org.apache.sysml.runtime.controlprogram.parfor.DataPartitioner.createPartitionedMatrixObject(DataPartitioner.java:162)
at org.apache.sysml.runtime.controlprogram.parfor.DataPartitioner.createPartitionedMatrixObject(DataPartitioner.java:81)
at org.apache.sysml.runtime.controlprogram.parfor.DataPartitioner.createPartitionedMatrixObject(DataPartitioner.java:69)
at org.apache.sysml.runtime.controlprogram.ParForProgramBlock.handleDataPartitioning(ParForProgramBlock.java:1140)
at org.apache.sysml.runtime.controlprogram.ParForProgramBlock.execute(ParForProgramBlock.java:565)
at org.apache.sysml.runtime.controlprogram.Program.execute(Program.java:123)
{code}
was:
Generally, we allow that users force singlenode execution type but for example a remote_mr or remote_spark parfor execution. For parfor remote_spark, this currently fails with the following stracktrace due to a wrong execution context.
{code}
Caused by: java.lang.ClassCastException: org.apache.sysml.runtime.controlprogram.context.ExecutionContext cannot be cast to org.apache.sysml.runtime.controlprogram.context.SparkExecutionContext
at org.apache.sysml.runtime.controlprogram.parfor.DataPartitionerRemoteSpark.partitionMatrix(DataPartitionerRemoteSpark.java:66)
at org.apache.sysml.runtime.controlprogram.parfor.DataPartitioner.createPartitionedMatrixObject(DataPartitioner.java:162)
at org.apache.sysml.runtime.controlprogram.parfor.DataPartitioner.createPartitionedMatrixObject(DataPartitioner.java:81)
at org.apache.sysml.runtime.controlprogram.parfor.DataPartitioner.createPartitionedMatrixObject(DataPartitioner.java:69)
at org.apache.sysml.runtime.controlprogram.ParForProgramBlock.handleDataPartitioning(ParForProgramBlock.java:1140)
at org.apache.sysml.runtime.controlprogram.ParForProgramBlock.execute(ParForProgramBlock.java:565)
at org.apache.sysml.runtime.controlprogram.Program.execute(Program.java:123)
{code}
> ClassCastException due to wrong execution context w/ parfor spark and forced singlenode
> ---------------------------------------------------------------------------------------
>
> Key: SYSTEMML-1386
> URL: https://issues.apache.org/jira/browse/SYSTEMML-1386
> Project: SystemML
> Issue Type: Bug
> Reporter: Matthias Boehm
>
> Generally, we allow that users force singlenode execution (of operations) but a remote_mr or remote_spark for parfor loops. For parfor remote_spark, this currently fails with the following stracktrace due to a wrong execution context.
> {code}
> Caused by: java.lang.ClassCastException: org.apache.sysml.runtime.controlprogram.context.ExecutionContext cannot be cast to org.apache.sysml.runtime.controlprogram.context.SparkExecutionContext
> at org.apache.sysml.runtime.controlprogram.parfor.DataPartitionerRemoteSpark.partitionMatrix(DataPartitionerRemoteSpark.java:66)
> at org.apache.sysml.runtime.controlprogram.parfor.DataPartitioner.createPartitionedMatrixObject(DataPartitioner.java:162)
> at org.apache.sysml.runtime.controlprogram.parfor.DataPartitioner.createPartitionedMatrixObject(DataPartitioner.java:81)
> at org.apache.sysml.runtime.controlprogram.parfor.DataPartitioner.createPartitionedMatrixObject(DataPartitioner.java:69)
> at org.apache.sysml.runtime.controlprogram.ParForProgramBlock.handleDataPartitioning(ParForProgramBlock.java:1140)
> at org.apache.sysml.runtime.controlprogram.ParForProgramBlock.execute(ParForProgramBlock.java:565)
> at org.apache.sysml.runtime.controlprogram.Program.execute(Program.java:123)
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)