Bobby Wang created SPARK-51537:
----------------------------------
Summary: Failed to run third-party Spark ML library on Spark
Connect
Key: SPARK-51537
URL: https://issues.apache.org/jira/browse/SPARK-51537
Project: Spark
Issue Type: Bug
Components: Connect, ML
Affects Versions: 4.0.0, 4.1
Reporter: Bobby Wang
I've encountered an issue where the third-party Spark ML library may not run
on Spark Connect. This problem occurs when specifying the
third-party ML jar using the *--jars* configuration while creating a connect
server
based on a Spark standalone cluster.
The exception thrown is a ClassCastException:
_Caused by: java.lang.ClassCastException: cannot assign instance of
java.lang.invoke.SerializedLambda to field
org.apache.spark.rdd.MapPartitionsRDD.f of type scala.Function3 in instance of
org.apache.spark.rdd.MapPartitionsRDD_
_at
java.io.ObjectStreamClass$FieldReflector.setObjFieldValues(ObjectStreamClass.java:2096)_
However, if I place the ML jar into the *$SPARK_HOME/jars* directory and
restart both the Spark standalone cluster and the Spark Connect server, it runs
without any exceptions.
Alternatively, adding
*spark.addArtifacts("target/com.example.ml-1.0-SNAPSHOT.jar")* directly in the
python code also resolves the issue.
I have made a minimum project which can repro this issue, more details could be
found at [https://github.com/wbo4958/ConnectMLIssue]
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
