[ https://issues.apache.org/jira/browse/SYSTEMML-1588?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Matthias Boehm reassigned SYSTEMML-1588: ---------------------------------------- Assignee: Matthias Boehm > Perftest: Datagen descriptive statistics fails on parfor result merge > --------------------------------------------------------------------- > > Key: SYSTEMML-1588 > URL: https://issues.apache.org/jira/browse/SYSTEMML-1588 > Project: SystemML > Issue Type: Bug > Affects Versions: SystemML 0.14 > Reporter: Matthias Boehm > Assignee: Matthias Boehm > Fix For: SystemML 1.0 > > > The perftest data generation script for descriptive statistics fails for the > 80GB scenario with the following stacktrace > {code} > Caused by: org.apache.sysml.runtime.DMLRuntimeException: > org.apache.sysml.runtime.DMLRuntimeException: > org.apache.hadoop.mapred.InvalidInputException: Input path does not exist: > hdfs://<node>:<port>/user/biuser/scratch_space/_p123456789_1.2.34.56/_t16/temp863_830 > at > org.apache.sysml.runtime.controlprogram.parfor.ResultMergeRemoteSpark.executeParallelMerge(ResultMergeRemoteSpark.java:116) > at > org.apache.sysml.runtime.controlprogram.parfor.ResultMergeRemoteSpark.executeSerialMerge(ResultMergeRemoteSpark.java:71) > at > org.apache.sysml.runtime.controlprogram.ParForProgramBlock.consolidateAndCheckResults(ParForProgramBlock.java:1756) > at > org.apache.sysml.runtime.controlprogram.ParForProgramBlock.executeLocalParFor(ParForProgramBlock.java:818) > at > org.apache.sysml.runtime.controlprogram.ParForProgramBlock.execute(ParForProgramBlock.java:639) > ... 14 more > {code} > This issue is caused by incorrect export handling of inputs which only works > for existing inputs on hdfs or dirty in-memory objects but not for pending > rdd operations. -- This message was sent by Atlassian JIRA (v6.3.15#6346)