spark-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Joe Mudd (JIRA)" <>
Subject [jira] [Commented] (SPARK-5435) saveAsNewAPIHadoopDataset is not setting up the local configuration
Date Fri, 08 Jan 2016 12:21:39 GMT


Joe Mudd commented on SPARK-5435:

I looked at the Hive 1.2.1 FileOutputFormatContainer implementation and it still references
org.apache.hadoop.mapred.FileOutputFormat.getUniqueName().  I'm not sure if this is a required
interop issue inherent to MapReduce.  Or, just the Hive folks being a bit slow to remove their
MRv1 references.

> saveAsNewAPIHadoopDataset is not setting up the local configuration
> -------------------------------------------------------------------
>                 Key: SPARK-5435
>                 URL:
>             Project: Spark
>          Issue Type: Bug
>          Components: Input/Output
>    Affects Versions: 1.2.0
>         Environment: Cloudera 5.3.0
>            Reporter: Joe Mudd
> The HCatOutputFormat utilizes FileOutpuFormatContainer which refers to the MRv1 FileOutputFormat.getUniqueName()
method.  Since the local configuration has not been set up, getUniqueName() ends up throwing
an IllegalArgumentException.
> It appears the saveAsNewAPIHadoopDataset().writeshard method needs to record Job information
in the local Hadoop configuration similar to HadoopRDD.addLocalConfiguration().  In a test
build, I ended up setting both the MRv1 and MRv2 names since just having the MRv2 names did
not work.
> Here's the traceback:
> java.lang.IllegalArgumentException: This method can only be called from within a Job
> 	at org.apache.hadoop.mapred.FileOutputFormat.getUniqueName(
> 	at org.apache.hive.hcatalog.mapreduce.FileOutputFormatContainer.getRecordWriter(
> 	at org.apache.hive.hcatalog.mapreduce.HCatOutputFormat.getRecordWriter(
> 	at org.apache.spark.rdd.PairRDDFunctions$$anonfun$12.apply(PairRDDFunctions.scala:984)
> 	at org.apache.spark.rdd.PairRDDFunctions$$anonfun$12.apply(PairRDDFunctions.scala:965)
> 	at org.apache.spark.scheduler.ResultTask.runTask(ResultTask.scala:61)
> 	at
> 	at org.apache.spark.executor.Executor$
> 	at java.util.concurrent.ThreadPoolExecutor.runWorker(
> 	at java.util.concurrent.ThreadPoolExecutor$

This message was sent by Atlassian JIRA

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message