Databricks Community

letsflykite · ‎07-31-2015

when I join two dataframes, I got the following error.

org.apache.spark.SparkException: Kryo serialization failed: Buffer overflow. Available: 0, required: 1 Serialization trace: values (org.apache.spark.sql.catalyst.expressions.GenericRow) otherElements (org.apache.spark.util.collection.CompactBuffer). To avoid this, increase spark.kryoserializer.buffer.max value. at org.apache.spark.serializer.KryoSerializerInstance.serialize(KryoSerializer.scala:253) at org.apache.spark.sql.execution.SparkSqlSerializer$$anonfun$serialize$1.apply(SparkSqlSerializer.scala:90) at org.apache.spark.sql.execution.SparkSqlSerializer$$anonfun$serialize$1.apply(SparkSqlSerializer.scala:89) at org.apache.spark.sql.execution.SparkSqlSerializer$.acquireRelease(SparkSqlSerializer.scala:82) at org.apache.spark.sql.execution.SparkSqlSerializer$.serialize(SparkSqlSerializer.scala:89) at org.apache.spark.sql.execution.joins.GeneralHashedRelation.writeExternal(HashedRelation.scala:65) at java.io.ObjectOutputStream.writeExternalData(ObjectOutputStream.java:1458) at java.io.ObjectOutputStream.writeOrdinaryObject(ObjectOutputStream.java:1429) at java.io.ObjectOutputStream.writeObject0(ObjectOutputStream.java:1177) at java.io.ObjectOutputStream.writeObject(ObjectOutputStream.java:347) at org.apache.spark.serializer.JavaSerializationStream.writeObject(JavaSerializer.scala:44) at org.apache.spark.broadcast.TorrentBroadcast$.blockifyObject(TorrentBroadcast.scala:203) at org.apache.spark.broadcast.TorrentBroadcast.writeBlocks(TorrentBroadcast.scala:102) at org.apache.spark.broadcast.TorrentBroadcast.<init>(TorrentBroadcast.scala:85) at org.apache.spark.broadcast.TorrentBroadcastFactory.newBroadcast(TorrentBroadcastFactory.scala:34) at org.apache.spark.broadcast.BroadcastManager.newBroadcast(BroadcastManager.scala:62) at org.apache.spark.SparkContext.broadcast(SparkContext.scala:1289)

So how to increase spark.kryoserializer.buffer.max in databricks cloud? http://spark.apache.org/docs/latest/configuration.html does not teach a way for databricks cloud.

arsalan1 · ‎08-07-2015

@letsflykite If you go to Databricks Guide -> Spark -> Configuring Spark you'll see a guide on how to change some of the Spark configuration settings using init scripts. On the near term roadmap will also be the ability to do these through the UI in an easier fashion.

One word of caution - it should be fairly rare to need to change these settings. Typically it means there is something in the code that is not performing as expected and leading to the error.