Skip to content

serializeToBundle issue #35

@habilmohammed

Description

@habilmohammed

Hi,
I am trying to serialize a simple spark model using mleap serializeToBundle method, but for some reason its failing with the below attached error, that the spark version is not supported. I have tried all the combinations of mleap and spark version as mentioned in the mleap doc, but the error remains the same.

Any help/suggestions on the resolving this would be appreciated.

python: 3.8
spark version 3.3.0
mleap: 0.22.0

data = spark.read.format("libsvm").load("sample_libsvm_data.txt")
featureIndexer = VectorIndexer(inputCol="features", outputCol="indexedFeatures", maxCategories=4).fit(data)
(trainingData, testData) = data.randomSplit([0.7, 0.3])
dt = DecisionTreeRegressor(featuresCol="indexedFeatures")
pipeline = Pipeline(stages=[featureIndexer, dt])
model = pipeline.fit(trainingData)

model.serializeToBundle("jar:file:/pyspark.example.zip", model.transform(data))
2024-07-22 17:29:21.070 IST
: java.lang.ExceptionInInitializerError
2024-07-22 17:29:21.070 IST
	at ml.combust.mleap.spark.SimpleSparkSerializer$$anonfun$1.apply(SimpleSparkSerializer.scala:17)
2024-07-22 17:29:21.070 IST
	at ml.combust.mleap.spark.SimpleSparkSerializer$$anonfun$1.apply(SimpleSparkSerializer.scala:17)
2024-07-22 17:29:21.070 IST
	at scala.Option.map(Option.scala:230)
2024-07-22 17:29:21.070 IST
	at ml.combust.mleap.spark.SimpleSparkSerializer.serializeToBundle(SimpleSparkSerializer.scala:17)
2024-07-22 17:29:21.070 IST
	at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
2024-07-22 17:29:21.070 IST
	at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
2024-07-22 17:29:21.070 IST
	at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
2024-07-22 17:29:21.070 IST
	at java.lang.reflect.Method.invoke(Method.java:498)
2024-07-22 17:29:21.070 IST
	at py4j.reflection.MethodInvoker.invoke(MethodInvoker.java:244)
2024-07-22 17:29:21.070 IST
	at py4j.reflection.ReflectionEngine.invoke(ReflectionEngine.java:357)
2024-07-22 17:29:21.070 IST
	at py4j.Gateway.invoke(Gateway.java:282)
2024-07-22 17:29:21.070 IST
	at py4j.commands.AbstractCommand.invokeMethod(AbstractCommand.java:132)
2024-07-22 17:29:21.070 IST
	at py4j.commands.CallCommand.execute(CallCommand.java:79)
2024-07-22 17:29:21.070 IST
	at py4j.ClientServerConnection.waitForCommands(ClientServerConnection.java:182)
2024-07-22 17:29:21.070 IST
	at py4j.ClientServerConnection.run(ClientServerConnection.java:106)
2024-07-22 17:29:21.070 IST
	at java.lang.Thread.run(Thread.java:750)
2024-07-22 17:29:21.070 IST
Caused by: java.lang.IllegalStateException: unsupported Spark version: 3.3.0
	at org.apache.spark.ml.bundle.SparkBundleContext$.<init>(SparkBundleContext.scala:23)
	at org.apache.spark.ml.bundle.SparkBundleContext$.<clinit>(SparkBundleContext.scala)
	... 16 more

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions