spark-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Tim Kellogg (Jira)" <j...@apache.org>
Subject [jira] [Comment Edited] (SPARK-30063) Failure when returning a value from multiple Pandas UDFs
Date Mon, 02 Dec 2019 23:09:00 GMT

    [ https://issues.apache.org/jira/browse/SPARK-30063?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16986438#comment-16986438
] 

Tim Kellogg edited comment on SPARK-30063 at 12/2/19 11:08 PM:
---------------------------------------------------------------

Additional notes on UDFs returning {{ArrayType(DoubleType())}}:
 * In pyarrow==0.8.0, the UDF return type must be a Python List. Returning pd.Series() will
crash.
 * In pyarrow==0.14.1, you can return pd.Series() also, it'll recognize it as a Spark Array.
Also, (IIRC) it'll coerce Array(FloatType()) to Array(DoubleType()) correctly.
 * In pyarrow==0.15.1 (latest), everything seems to break again, I believe it's due to a change
in message format, introducing that leading -1 byte.

So in short, pin to pyarrow==0.8.0 and coerce all arrays to native Python list before returning
from the UDF.


was (Author: tkellogg):
Additional notes on UDFs returning {{ArrayType(DoubleType())}}:
 * In pyarrow==0.8.0, the UDF return type must be a Python List. Returning pd.Series() will
crash.
 * In pyarrow==0.14.1, you can return pd.Series() also, it'll recognize it as a Spark Array
 * In pyarrow==0.15.1 (latest), everything seems to break again, I believe it's due to a change
in message format, introducing that leading -1 byte.

So in short, pin to pyarrow==0.8.0 and coerce all arrays to native Python list before returning
from the UDF.

> Failure when returning a value from multiple Pandas UDFs
> --------------------------------------------------------
>
>                 Key: SPARK-30063
>                 URL: https://issues.apache.org/jira/browse/SPARK-30063
>             Project: Spark
>          Issue Type: Bug
>          Components: PySpark
>    Affects Versions: 2.4.3, 2.4.4
>         Environment: Happens on Mac & Ubuntu (Docker). Seems to happen on both 2.4.3
and 2.4.4
>            Reporter: Tim Kellogg
>            Priority: Major
>         Attachments: spark-debug.txt, variety-of-schemas.ipynb
>
>
> I have 20 Pandas UDFs that I'm trying to evaluate all at the same time.
>  * PandasUDFType.GROUPED_AGG
>  * 3 columns in the input data frame being serialized over Arrow to Python worker. See
below for clarification.
>  * All functions take 2 parameters, some combination of the 3 received as Arrow input.
>  * Varying return types, see details below.
> _*I get an IllegalArgumentException on the Scala side of the worker when deserializing
from Python.*_
> h2. Exception & Stack Trace
> {code:java}
> 19/11/27 11:38:36 ERROR Executor: Exception in task 0.0 in stage 5.0 (TID 5)
> java.lang.IllegalArgumentException
> 	at java.nio.ByteBuffer.allocate(ByteBuffer.java:334)
> 	at org.apache.arrow.vector.ipc.message.MessageSerializer.readMessage(MessageSerializer.java:543)
> 	at org.apache.arrow.vector.ipc.message.MessageChannelReader.readNext(MessageChannelReader.java:58)
> 	at org.apache.arrow.vector.ipc.ArrowStreamReader.readSchema(ArrowStreamReader.java:132)
> 	at org.apache.arrow.vector.ipc.ArrowReader.initialize(ArrowReader.java:181)
> 	at org.apache.arrow.vector.ipc.ArrowReader.ensureInitialized(ArrowReader.java:172)
> 	at org.apache.arrow.vector.ipc.ArrowReader.getVectorSchemaRoot(ArrowReader.java:65)
> 	at org.apache.spark.sql.execution.python.ArrowPythonRunner$$anon$1.read(ArrowPythonRunner.scala:162)
> 	at org.apache.spark.sql.execution.python.ArrowPythonRunner$$anon$1.read(ArrowPythonRunner.scala:122)
> 	at org.apache.spark.api.python.BasePythonRunner$ReaderIterator.hasNext(PythonRunner.scala:410)
> 	at org.apache.spark.InterruptibleIterator.hasNext(InterruptibleIterator.scala:37)
> 	at scala.collection.Iterator$$anon$11.hasNext(Iterator.scala:409)
> 	at scala.collection.Iterator$$anon$11.hasNext(Iterator.scala:409)
> 	at org.apache.spark.sql.execution.SparkPlan$$anonfun$2.apply(SparkPlan.scala:255)
> 	at org.apache.spark.sql.execution.SparkPlan$$anonfun$2.apply(SparkPlan.scala:247)
> 	at org.apache.spark.rdd.RDD$$anonfun$mapPartitionsInternal$1$$anonfun$apply$24.apply(RDD.scala:836)
> 	at org.apache.spark.rdd.RDD$$anonfun$mapPartitionsInternal$1$$anonfun$apply$24.apply(RDD.scala:836)
> 	at org.apache.spark.rdd.MapPartitionsRDD.compute(MapPartitionsRDD.scala:52)
> 	at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:324)
> 	at org.apache.spark.rdd.RDD.iterator(RDD.scala:288)
> 	at org.apache.spark.scheduler.ResultTask.runTask(ResultTask.scala:90)
> 	at org.apache.spark.scheduler.Task.run(Task.scala:123)
> 	at org.apache.spark.executor.Executor$TaskRunner$$anonfun$10.apply(Executor.scala:408)
> 	at org.apache.spark.util.Utils$.tryWithSafeFinally(Utils.scala:1360)
> 	at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:414)
> 	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
> 	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
> 	at java.lang.Thread.run(Thread.java:748)
> 19/11/27 11:38:36 WARN TaskSetManager: Lost task 0.0 in stage 5.0 (TID 5, localhost,
executor driver): java.lang.IllegalArgumentException
> 	at java.nio.ByteBuffer.allocate(ByteBuffer.java:334)
> 	at org.apache.arrow.vector.ipc.message.MessageSerializer.readMessage(MessageSerializer.java:543)
> 	at org.apache.arrow.vector.ipc.message.MessageChannelReader.readNext(MessageChannelReader.java:58)
> 	at org.apache.arrow.vector.ipc.ArrowStreamReader.readSchema(ArrowStreamReader.java:132)
> 	at org.apache.arrow.vector.ipc.ArrowReader.initialize(ArrowReader.java:181)
> 	at org.apache.arrow.vector.ipc.ArrowReader.ensureInitialized(ArrowReader.java:172)
> 	at org.apache.arrow.vector.ipc.ArrowReader.getVectorSchemaRoot(ArrowReader.java:65)
> 	at org.apache.spark.sql.execution.python.ArrowPythonRunner$$anon$1.read(ArrowPythonRunner.scala:162)
> 	at org.apache.spark.sql.execution.python.ArrowPythonRunner$$anon$1.read(ArrowPythonRunner.scala:122)
> 	at org.apache.spark.api.python.BasePythonRunner$ReaderIterator.hasNext(PythonRunner.scala:410)
> 	at org.apache.spark.InterruptibleIterator.hasNext(InterruptibleIterator.scala:37)
> 	at scala.collection.Iterator$$anon$11.hasNext(Iterator.scala:409)
> 	at scala.collection.Iterator$$anon$11.hasNext(Iterator.scala:409)
> 	at org.apache.spark.sql.execution.SparkPlan$$anonfun$2.apply(SparkPlan.scala:255)
> 	at org.apache.spark.sql.execution.SparkPlan$$anonfun$2.apply(SparkPlan.scala:247)
> 	at org.apache.spark.rdd.RDD$$anonfun$mapPartitionsInternal$1$$anonfun$apply$24.apply(RDD.scala:836)
> 	at org.apache.spark.rdd.RDD$$anonfun$mapPartitionsInternal$1$$anonfun$apply$24.apply(RDD.scala:836)
> 	at org.apache.spark.rdd.MapPartitionsRDD.compute(MapPartitionsRDD.scala:52)
> 	at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:324)
> 	at org.apache.spark.rdd.RDD.iterator(RDD.scala:288)
> 	at org.apache.spark.scheduler.ResultTask.runTask(ResultTask.scala:90)
> 	at org.apache.spark.scheduler.Task.run(Task.scala:123)
> 	at org.apache.spark.executor.Executor$TaskRunner$$anonfun$10.apply(Executor.scala:408)
> 	at org.apache.spark.util.Utils$.tryWithSafeFinally(Utils.scala:1360)
> 	at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:414)
> 	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
> 	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
> 	at java.lang.Thread.run(Thread.java:748)
> {code}
> h2. Input Arrow Schema
> I edited ArrowStreamPandasSerializer in pyspark/serializers.py to print out schema &
message. This is the input, in load_stream, the code is print(batch, batch.schema, file=log_file)
> {code:java}
> <pyarrow.lib.RecordBatch object at 0x10640ecc8> 
> _0: double
> _1: double
> _2: double
> metadata
> --------
> OrderedDict()
> {code}
> h2. Output Arrow Schema
> I edited ArrowStreamPandasSerializer in pyspark/serializers.py to print out schema &
message. This is the output, in dump_stream, the code is print(batch, batch.schema, file=log_file)
> {code:java}
> <pyarrow.lib.RecordBatch object at 0x11ad5b638> _0: float
> _1: float
> _2: float
> _3: int32
> _4: int32
> _5: int32
> _6: int32
> _7: int32
> _8: float
> _9: float
> _10: int32
> _11: list<item: float>
>   child 0, item: float
> _12: list<item: float>
>   child 0, item: float
> _13: float
> _14: float
> _15: int32
> _16: float
> _17: list<item: float>
>   child 0, item: float
> _18: list<item: float>
>   child 0, item: float
> _19: float
> {code}
> h2. Arrow Message
> I edited ArrowPythonReader.scala at line 163 to print out the Arrow message.
> Debug code:
> {code:java}
> val fw = new java.io.FileWriter("spark-debug.txt", true)
> try {
>   val buf = new Array[Byte](40000)
>   stream.read(buf)
>   fw.write(s"Spark reader\n")
>   for (b <- buf) {
>     fw.write(String.format("%02x", Byte.box(b)))
>   }
>   fw.write(s"\n")
> } finally fw.close()
> {code}
> Debug output (some trailing 0's included for completeness).
> {code:java}
> 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
> {code}
>  
> h2. Query Plan at Point of Failure
> Right before the failure, I printed out the explain(True) output.
>  
> {code:java}
> == Parsed Logical Plan ==
> 'Project [structstojson(named_struct(), None) AS key#269, unresolvedalias('accuracy,
None), unresolvedalias('areaUnderPR, None), unresolvedalias('areaUnderROC, None), unresolvedalias('confusionMatrix,
None), unresolvedalias('count, None), unresolvedalias('f1Score, None), unresolvedalias('f1Score_0,
None), unresolvedalias('positiveClassRate, None), unresolvedalias('prCurve, None), unresolvedalias('precision,
None), unresolvedalias('precision_0, None), unresolvedalias('predictionRate, None), unresolvedalias('recall,
None), unresolvedalias('rocCurve, None), unresolvedalias('specificity, None)]
> +- Aggregate [udf(cast(label#146 as double), cast(prediction#15 as double)) AS accuracy#232,
_auc_pr(cast(label#146 as double), cast(probability#16 as double)) AS areaUnderPR#233, udf(cast(label#146
as double), cast(probability#16 as double)) AS areaUnderROC#225, array(array(udf(cast(label#146
as double), cast(prediction#15 as double)), udf(cast(label#146 as double), cast(prediction#15
as double))), array(udf(cast(label#146 as double), cast(prediction#15 as double)), udf(cast(label#146
as double), cast(prediction#15 as double)))) AS confusionMatrix#238, _count(cast(label#146
as double)) AS count#234, udf(cast(label#146 as double), cast(prediction#15 as double)) AS
f1Score#231, udf(cast(label#146 as double), cast(prediction#15 as double)) AS f1Score_0#228,
_rate(cast(label#146 as double)) AS positiveClassRate#227, named_struct(x, udf(cast(label#146
as double), cast(probability#16 as double)), y, udf(cast(label#146 as double), cast(probability#16
as double))) AS prCurve#230, udf(cast(label#146 as double), cast(prediction#15 as double))
AS precision#236, udf(cast(label#146 as double), cast(prediction#15 as double)) AS precision_0#235,
_rate(cast(prediction#15 as double)) AS predictionRate#237, udf(cast(label#146 as double),
cast(prediction#15 as double)) AS recall#229, named_struct(x, udf(cast(label#146 as double),
cast(probability#16 as double)), y, udf(cast(label#146 as double), cast(probability#16 as
double))) AS rocCurve#239, udf(cast(label#146 as double), cast(prediction#15 as double)) AS
specificity#226]
>    +- Project [encounterID#13, dim1#11, dim2#12, prediction#15, probability#16, model_id#23,
label#146, test AS customer#156, foo AS solution#157, bar AS insight#158, model AS model_name#159,
1.0 AS version#160, model1 AS model_id#161, current_timestamp() AS timestamp#162]
>       +- Project [encounterID#13, dim1#11, dim2#12, prediction#15, probability#16, model_id#23,
label#146]
>          +- Join Inner, (encounterID#13 = encounterID#145)
>             :- Project [dim1#11, dim2#12, encounterID#13, prediction#15, probability#16,
model_id#23]
>             :  +- Filter ((false || NOT test#40) = false)
>             :     +- Project [dim1#11, dim2#12, encounterID#13, prediction#15, probability#16,
model_id#23, (true && (dim1#11 = foo)) AS test#40]
>             :        +- Project [dim1#11, dim2#12, encounterID#13, prediction#15, probability#16,
model_id#23]
>             :           +- Join Cross
>             :              :- LogicalRDD [dim1#11, dim2#12, encounterID#13, label#14,
prediction#15, probability#16], false
>             :              +- LogicalRDD [model_id#23], false
>             +- Project [encounterID#145, label#146]
>                +- Join Cross
>                   :- LogicalRDD [dim1#143, dim2#144, encounterID#145, label#146, prediction#147,
probability#148], false
>                   +- LogicalRDD [model_id#23], false== Analyzed Logical Plan ==
> key: string, accuracy: float, areaUnderPR: float, areaUnderROC: float, confusionMatrix:
array<array<int>>, count: int, f1Score: float, f1Score_0: float, positiveClassRate:
int, prCurve: struct<x:array<float>,y:array<float>>, precision: float, precision_0:
float, predictionRate: int, recall: float, rocCurve: struct<x:array<float>,y:array<float>>,
specificity: float
> Project [structstojson(named_struct(), Some(America/Los_Angeles)) AS key#269, accuracy#232,
areaUnderPR#233, areaUnderROC#225, confusionMatrix#238, count#234, f1Score#231, f1Score_0#228,
positiveClassRate#227, prCurve#230, precision#236, precision_0#235, predictionRate#237, recall#229,
rocCurve#239, specificity#226]
> +- Aggregate [udf(cast(label#146 as double), cast(prediction#15 as double)) AS accuracy#232,
_auc_pr(cast(label#146 as double), cast(probability#16 as double)) AS areaUnderPR#233, udf(cast(label#146
as double), cast(probability#16 as double)) AS areaUnderROC#225, array(array(udf(cast(label#146
as double), cast(prediction#15 as double)), udf(cast(label#146 as double), cast(prediction#15
as double))), array(udf(cast(label#146 as double), cast(prediction#15 as double)), udf(cast(label#146
as double), cast(prediction#15 as double)))) AS confusionMatrix#238, _count(cast(label#146
as double)) AS count#234, udf(cast(label#146 as double), cast(prediction#15 as double)) AS
f1Score#231, udf(cast(label#146 as double), cast(prediction#15 as double)) AS f1Score_0#228,
_rate(cast(label#146 as double)) AS positiveClassRate#227, named_struct(x, udf(cast(label#146
as double), cast(probability#16 as double)), y, udf(cast(label#146 as double), cast(probability#16
as double))) AS prCurve#230, udf(cast(label#146 as double), cast(prediction#15 as double))
AS precision#236, udf(cast(label#146 as double), cast(prediction#15 as double)) AS precision_0#235,
_rate(cast(prediction#15 as double)) AS predictionRate#237, udf(cast(label#146 as double),
cast(prediction#15 as double)) AS recall#229, named_struct(x, udf(cast(label#146 as double),
cast(probability#16 as double)), y, udf(cast(label#146 as double), cast(probability#16 as
double))) AS rocCurve#239, udf(cast(label#146 as double), cast(prediction#15 as double)) AS
specificity#226]
>    +- Project [encounterID#13, dim1#11, dim2#12, prediction#15, probability#16, model_id#23,
label#146, test AS customer#156, foo AS solution#157, bar AS insight#158, model AS model_name#159,
1.0 AS version#160, model1 AS model_id#161, current_timestamp() AS timestamp#162]
>       +- Project [encounterID#13, dim1#11, dim2#12, prediction#15, probability#16, model_id#23,
label#146]
>          +- Join Inner, (encounterID#13 = encounterID#145)
>             :- Project [dim1#11, dim2#12, encounterID#13, prediction#15, probability#16,
model_id#23]
>             :  +- Filter ((false || NOT test#40) = false)
>             :     +- Project [dim1#11, dim2#12, encounterID#13, prediction#15, probability#16,
model_id#23, (true && (dim1#11 = foo)) AS test#40]
>             :        +- Project [dim1#11, dim2#12, encounterID#13, prediction#15, probability#16,
model_id#23]
>             :           +- Join Cross
>             :              :- LogicalRDD [dim1#11, dim2#12, encounterID#13, label#14,
prediction#15, probability#16], false
>             :              +- LogicalRDD [model_id#23], false
>             +- Project [encounterID#145, label#146]
>                +- Join Cross
>                   :- LogicalRDD [dim1#143, dim2#144, encounterID#145, label#146, prediction#147,
probability#148], false
>                   +- LogicalRDD [model_id#23], false== Optimized Logical Plan ==
> Aggregate [{} AS key#269, udf(label#146, prediction#15) AS accuracy#232, _auc_pr(label#146,
probability#16) AS areaUnderPR#233, udf(label#146, probability#16) AS areaUnderROC#225, array(array(udf(label#146,
prediction#15), udf(label#146, prediction#15)), array(udf(label#146, prediction#15), udf(label#146,
prediction#15))) AS confusionMatrix#238, _count(label#146) AS count#234, udf(label#146, prediction#15)
AS f1Score#231, udf(label#146, prediction#15) AS f1Score_0#228, _rate(label#146) AS positiveClassRate#227,
named_struct(x, udf(label#146, probability#16), y, udf(label#146, probability#16)) AS prCurve#230,
udf(label#146, prediction#15) AS precision#236, udf(label#146, prediction#15) AS precision_0#235,
_rate(prediction#15) AS predictionRate#237, udf(label#146, prediction#15) AS recall#229, named_struct(x,
udf(label#146, probability#16), y, udf(label#146, probability#16)) AS rocCurve#239, udf(label#146,
prediction#15) AS specificity#226]
> +- Project [prediction#15, probability#16, label#146]
>    +- Join Inner, (encounterID#13 = encounterID#145)
>       :- Project [encounterID#13, prediction#15, probability#16]
>       :  +- Filter ((isnotnull(test#40) && (NOT test#40 = false)) &&
isnotnull(encounterID#13))
>       :     +- InMemoryRelation [dim1#11, dim2#12, encounterID#13, prediction#15, probability#16,
model_id#23, test#40], StorageLevel(disk, memory, deserialized, 1 replicas)
>       :           +- *(2) Project [dim1#11, dim2#12, encounterID#13, prediction#15, probability#16,
model_id#23, (dim1#11 = foo) AS test#40]
>       :              +- CartesianProduct
>       :                 :- *(1) Project [dim1#11, dim2#12, encounterID#13, prediction#15,
probability#16]
>       :                 :  +- Scan ExistingRDD[dim1#11,dim2#12,encounterID#13,label#14,prediction#15,probability#16]
>       :                 +- Scan ExistingRDD[model_id#23]
>       +- Join Cross
>          :- Project [encounterID#145, label#146]
>          :  +- Filter isnotnull(encounterID#145)
>          :     +- LogicalRDD [dim1#143, dim2#144, encounterID#145, label#146, prediction#147,
probability#148], false
>          +- Project
>             +- LogicalRDD [model_id#23], false== Physical Plan ==
> !AggregateInPandas [udf(label#146, prediction#15), _auc_pr(label#146, probability#16),
udf(label#146, probability#16), udf(label#146, prediction#15), udf(label#146, prediction#15),
udf(label#146, prediction#15), udf(label#146, prediction#15), _count(label#146), udf(label#146,
prediction#15), udf(label#146, prediction#15), _rate(label#146), udf(label#146, probability#16),
udf(label#146, probability#16), udf(label#146, prediction#15), udf(label#146, prediction#15),
_rate(prediction#15), udf(label#146, prediction#15), udf(label#146, probability#16), udf(label#146,
probability#16), udf(label#146, prediction#15)], [{} AS key#269, udf(label, prediction)#201
AS accuracy#232, _auc_pr(label, probability)#209 AS areaUnderPR#233, udf(label, probability)#208
AS areaUnderROC#225, array(array(udf(label, prediction)#213, udf(label, prediction)#214),
array(udf(label, prediction)#215, udf(label, prediction)#216)) AS confusionMatrix#238, _count(label)#210
AS count#234, udf(label, prediction)#206 AS f1Score#231, udf(label, prediction)#207 AS f1Score_0#228,
_rate(label)#212 AS positiveClassRate#227, named_struct(x, udf(label, probability)#219, y,
udf(label, probability)#220) AS prCurve#230, udf(label, prediction)#202 AS precision#236,
udf(label, prediction)#203 AS precision_0#235, _rate(prediction)#211 AS predictionRate#237,
udf(label, prediction)#204 AS recall#229, named_struct(x, udf(label, probability)#217, y,
udf(label, probability)#218) AS rocCurve#239, udf(label, prediction)#205 AS specificity#226]
> +- Exchange SinglePartition
>    +- *(4) Project [prediction#15, probability#16, label#146]
>       +- *(4) BroadcastHashJoin [encounterID#13], [encounterID#145], Inner, BuildLeft
>          :- BroadcastExchange HashedRelationBroadcastMode(List(input[0, string, true]))
>          :  +- *(1) Project [encounterID#13, prediction#15, probability#16]
>          :     +- *(1) Filter ((isnotnull(test#40) && (NOT test#40 = false))
&& isnotnull(encounterID#13))
>          :        +- InMemoryTableScan [encounterID#13, prediction#15, probability#16,
test#40], [isnotnull(test#40), (NOT test#40 = false), isnotnull(encounterID#13)]
>          :              +- InMemoryRelation [dim1#11, dim2#12, encounterID#13, prediction#15,
probability#16, model_id#23, test#40], StorageLevel(disk, memory, deserialized, 1 replicas)
>          :                    +- *(2) Project [dim1#11, dim2#12, encounterID#13, prediction#15,
probability#16, model_id#23, (dim1#11 = foo) AS test#40]
>          :                       +- CartesianProduct
>          :                          :- *(1) Project [dim1#11, dim2#12, encounterID#13,
prediction#15, probability#16]
>          :                          :  +- Scan ExistingRDD[dim1#11,dim2#12,encounterID#13,label#14,prediction#15,probability#16]
>          :                          +- Scan ExistingRDD[model_id#23]
>          +- CartesianProduct
>             :- *(2) Project [encounterID#145, label#146]
>             :  +- *(2) Filter isnotnull(encounterID#145)
>             :     +- Scan ExistingRDD[dim1#143,dim2#144,encounterID#145,label#146,prediction#147,probability#148]
>             +- *(3) Project
>                +- Scan ExistingRDD[model_id#23]
> {code}
> h2. Related Bugs
> I have a related bug that I've gotten where the schema in the input Arrow message was
transmiitted incorrectly. In that case, the input schema should have been <long, float,
long> but was transmitted as <long, long, float>. As a result, the float column was
interpreted as a long (equivalent C code to illustrate behavior: )
> {code:java}
> long reinterpret(double floating_point_number) {
>   return *(long*)(&floating_point_number)
> }
> {code}
> I got around this bug by making all 3 columns float and converting them to long within
the UDF via Pandas Series.apply(np.int). Strangely, a Column.astype('float') didn't seem to
have an effect, I had to make them float at the source.
> Along the way, I had trouble with [Python's dict keys being non-deterministic|[https://stackoverflow.com/questions/14956313/why-is-dictionary-ordering-non-deterministic].]
This led columns being passed to GroupedData.agg() in different orders for each worker and
driver process. I've mitigated this by explicitly ordering the columns before sending them
to agg. I don't think this is an issue anymore, but I'm calling it out just in case.
>  
>  



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org


Mime
View raw message