Are you registering multiple RDDs of case classes as tables concurrently? You are possibly hitting SPARK-2178 <https://issues.apache.org/jira/browse/SPARK-2178> which is caused by SI-6240 <https://issues.scala-lang.org/browse/SI-6240>.
On Tue, Jul 15, 2014 at 10:49 AM, Keith Simmons <[email protected]> wrote: > HI folks, > > I'm running into the following error when trying to perform a join in my > code: > > java.lang.NoClassDefFoundError: Could not initialize class > org.apache.spark.sql.catalyst.types.LongType$ > > I see similar errors for StringType$ and also: > > scala.reflect.runtime.ReflectError: value apache is not a package. > > Strangely, if I just work with a single table, everything is fine. I can > iterate through the records in both tables and print them out without a > problem. > > Furthermore, this code worked without an exception in Spark 1.0.0 (thought > the join caused some field corruption, possibly related to > https://issues.apache.org/jira/browse/SPARK-1994 > <https://www.google.com/url?q=https%3A%2F%2Fissues.apache.org%2Fjira%2Fbrowse%2FSPARK-1994&sa=D&sntz=1&usg=AFQjCNHNxePxWgmuymCQSprulDZZcOn4-Q>). > The data is coming from a custom protocol buffer based format on hdfs that > is being mapped into the individual record types without a problem. > > The immediate cause seems to be a task trying to deserialize one or more > SQL case classes before loading the spark uber jar, but I have no idea why > this is happening, or why it only happens when I do a join. Ideas? > > Keith > > P.S. If it's relevant, we're using the Kryo serializer. > > >
