spark-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Miao Wang (JIRA)" <j...@apache.org>
Subject [jira] [Comment Edited] (SPARK-18011) SparkR serialize "NA" throws exception
Date Tue, 17 Jan 2017 23:27:26 GMT

    [ https://issues.apache.org/jira/browse/SPARK-18011?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15827032#comment-15827032
] 

Miao Wang edited comment on SPARK-18011 at 1/17/17 11:27 PM:
-------------------------------------------------------------

[~felixcheung] I did intensive debug in both R side and scala side.

On R side, I debugged `createDataFrame.default` and `parallelize`, which converts the data.frame
into RDD and DataFrame. The code of turning the data into RDD is done in `parallelize`:
 sliceLen <- ceiling(length(coll) / numSlices)
 slices <- split(coll, rep(1: (numSlices + 1), each = sliceLen)[1:length(coll)])
 serializedSlices <- lapply(slices, serialize, connection = NULL)

I add debug message after the `serialize`:
lapply(serializedSlices, function(`x`) {message(paste("unserialized ", unserialize(x)))})

The data `NA` is unserialized successfully. 

Then, the serialized data is transferred to Scala side by jrdd <- callJStatic("org.apache.spark.api.r.RRDD",
"createRDDFromArray", sc, serializedSlices) and returns a handle of the RDD in `jrdd`, which
is later used by `createDataFrame.default`.

I did not find anything wrong here.

On the Scala side, the problem happens in 

def readString(in: DataInputStream): String = {
    val len = in.readInt() <=== it encounters the problem when reading `NA` as a string.
    readStringBytes(in, len)
  }

Then, I changed the logic as follows:
 def readString(in: DataInputStream): String = {
    var len = in.readInt()
    if (len < 0) {
      len = 3<===== I enforce reading 3 bytes in this case, because I believe that it is
the case of `NA`
    }
    readStringBytes(in, len)
  }

Then I run the following commands in sparkR:

> a <- as.Date(NA)
> b <- as.data.frame(a)
> c <- collect(select(createDataFrame(b), "*"))
> c
   a
1 NA
It executes correctly without hitting the exception handling (I add debug information in the
handling logic. If it is hit, error message will be print on the console and I verified that
it is print out without the above logic).

So, we can conclude that the problem is caused by `serialize` function with my local R installation,
which serialize `NA` as string without packing its length before the actual value. Since `unserialize`
can decode the seralized data, this protocol should be by R design when handling `NA` as `Date`
type. I don't find the source code of `serialize` in R source code, which calls Internal(serialize(object,
connection, type, version, refhook))

For the fix, we can either leave it as it is by an exception handling or explicitly add a
handling in readString when index is negative.

What do you think? Thanks!  



was (Author: wm624):
[~felixcheung] I did intensive debug in both R side and scala side.

On R side, I debugged `createDataFrame.default` and `parallelize`, which converts the data.frame
into RDD and DataFrame. The code of turning the data into RDD is done in `parallelize`:
 sliceLen <- ceiling(length(coll) / numSlices)
 slices <- split(coll, rep(1: (numSlices + 1), each = sliceLen)[1:length(coll)])
 serializedSlices <- lapply(slices, serialize, connection = NULL)

I add debug message after the `serialize`:
lapply(serializedSlices, function(x) {message(paste("unserialized ", unserialize(x)))})

The data `NA` is unserialized successfully. 

Then, the serialized data is transferred to Scala side by jrdd <- callJStatic("org.apache.spark.api.r.RRDD",
"createRDDFromArray", sc, serializedSlices) and returns a handle of the RDD in `jrdd`, which
is later used by `createDataFrame.default`.

I did not find anything wrong here.

On the Scala side, the problem happens in 

def readString(in: DataInputStream): String = {
    val len = in.readInt() <=== it encounters the problem when reading `NA` as a string.
    readStringBytes(in, len)
  }

Then, I changed the logic as follows:
 def readString(in: DataInputStream): String = {
    var len = in.readInt()
    if (len < 0) {
      len = 3<===== I enforce reading 3 bytes in this case, because I believe that it is
the case of `NA`
    }
    readStringBytes(in, len)
  }

Then I run the following commands in sparkR:

> a <- as.Date(NA)
> b <- as.data.frame(a)
> c <- collect(select(createDataFrame(b), "*"))
> c
   a
1 NA
It executes correctly without hitting the exception handling (I add debug information in the
handling logic. If it is hit, error message will be print on the console and I verified that
it is print out without the above logic).

So, we can conclude that the problem is caused by `serialize` function with my local R installation,
which serialize `NA` as string without packing its length before the actual value. Since `unserialize`
can decode the seralized data, this protocol should be by R design when handling `NA` as `Date`
type. I don't find the source code of `serialize` in R source code, which calls Internal(serialize(object,
connection, type, version, refhook))

For the fix, we can either leave it as it is by an exception handling or explicitly add a
handling in readString when index is negative.

What do you think? Thanks!  


> SparkR serialize "NA" throws exception
> --------------------------------------
>
>                 Key: SPARK-18011
>                 URL: https://issues.apache.org/jira/browse/SPARK-18011
>             Project: Spark
>          Issue Type: Bug
>          Components: SparkR
>            Reporter: Miao Wang
>
> For some versions of R, if Date has "NA" field, backend will throw negative index exception.
> To reproduce the problem:
> {code}
> > a <- as.Date(c("2016-11-11", "NA"))
> > b <- as.data.frame(a)
> > c <- createDataFrame(b)
> > dim(c)
> 16/10/19 10:31:24 ERROR Executor: Exception in task 0.0 in stage 1.0 (TID 1)
> java.lang.NegativeArraySizeException
> 	at org.apache.spark.api.r.SerDe$.readStringBytes(SerDe.scala:110)
> 	at org.apache.spark.api.r.SerDe$.readString(SerDe.scala:119)
> 	at org.apache.spark.api.r.SerDe$.readDate(SerDe.scala:128)
> 	at org.apache.spark.api.r.SerDe$.readTypedObject(SerDe.scala:77)
> 	at org.apache.spark.api.r.SerDe$.readObject(SerDe.scala:61)
> 	at org.apache.spark.sql.api.r.SQLUtils$$anonfun$bytesToRow$1.apply(SQLUtils.scala:161)
> 	at org.apache.spark.sql.api.r.SQLUtils$$anonfun$bytesToRow$1.apply(SQLUtils.scala:160)
> 	at scala.collection.TraversableLike$$anonfun$map$1.apply(TraversableLike.scala:234)
> 	at scala.collection.TraversableLike$$anonfun$map$1.apply(TraversableLike.scala:234)
> 	at scala.collection.immutable.Range.foreach(Range.scala:160)
> 	at scala.collection.TraversableLike$class.map(TraversableLike.scala:234)
> 	at scala.collection.AbstractTraversable.map(Traversable.scala:104)
> 	at org.apache.spark.sql.api.r.SQLUtils$.bytesToRow(SQLUtils.scala:160)
> 	at org.apache.spark.sql.api.r.SQLUtils$$anonfun$5.apply(SQLUtils.scala:138)
> 	at org.apache.spark.sql.api.r.SQLUtils$$anonfun$5.apply(SQLUtils.scala:138)
> 	at scala.collection.Iterator$$anon$11.next(Iterator.scala:409)
> 	at scala.collection.Iterator$$anon$11.next(Iterator.scala:409)
> 	at scala.collection.Iterator$$anon$11.next(Iterator.scala:409)
> 	at org.apache.spark.sql.catalyst.expressions.GeneratedClass$GeneratedIterator.agg_doAggregateWithoutKey$(Unknown
Source)
> 	at org.apache.spark.sql.catalyst.expressions.GeneratedClass$GeneratedIterator.processNext(Unknown
Source)
> 	at org.apache.spark.sql.execution.BufferedRowIterator.hasNext(BufferedRowIterator.java:43)
> 	at org.apache.spark.sql.execution.WholeStageCodegenExec$$anonfun$8$$anon$1.hasNext(WholeStageCodegenExec.scala:372)
> 	at scala.collection.Iterator$$anon$11.hasNext(Iterator.scala:408)
> 	at org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:126)
> 	at org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:96)
> 	at org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:53)
> 	at org.apache.spark.scheduler.Task.run(Task.scala:99)
> 	at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:282)
> 	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> 	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> 	at java.lang.Thread.run(Thread.java:745)
> {code}



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

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


Mime
View raw message