spark: SAXParseException while writing to parquet on s3 - scala

Question

I'm trying to read in some json, infer a schema, and write it out again as parquet to s3 (s3a). For some reason, about a third of the way through the writing portion of the run, spark always errors out with the error included below. I can't find any obvious reasons for the issue: it isn't out of memory; there are no long GC pauses. There don't seem to be any additional error messages in the logs of the individual executors.
The script runs fine on another set of data that I have, which is of a very similar structure, but several orders of magnitude smaller.
I am running spark 2.0.1-hadoop-2.7 and am using the FileOutputCommitter. The algorithm version doesn't seem to matter.
Edit:
This does not appear to be a problem in badly formed json or corrupted files. I have unzipped and read in each file individually with no error.
Here's a simplified version of the script:
object Foo {
def parseJson(json: String): Option[Map[String, Any]] = {
if (json == null)
Some(Map())
else
parseOpt(json).map((j: JValue) => j.values.asInstanceOf[Map[String, Any]])
}
}
}
// read in as text and parse json using json4s
val jsonRDD: RDD[String] = sc.textFile(inputPath)
.map(row -> Foo.parseJson(row))
// infer a schema that will encapsulate the most rows in a sample of size sampleRowNum
val schema: StructType = Infer.getMostCommonSchema(sc, jsonRDD, sampleRowNum)
// get documents compatibility with schema
val jsonWithCompatibilityRDD: RDD[(String, Boolean)] = jsonRDD
.map(js => (js, Infer.getSchemaCompatibility(schema, Infer.inferSchema(js)).toBoolean))
.repartition(partitions)
val jsonCompatibleRDD: RDD[String] = jsonWithCompatibilityRDD
.filter { case (js: String, compatible: Boolean) => compatible }
.map { case (js: String, _: Boolean) => js }
// create a dataframe from documents with compatible schema
val dataFrame: DataFrame = spark.read.schema(schema).json(jsonCompatibleRDD)
It completes the earlier schema inferring steps successfully. The error itself occurs on the last line, but I suppose that could encompass at least the immediately preceding statemnt, if not earlier:
org.apache.spark.SparkException: Task failed while writing rows
at org.apache.spark.sql.execution.datasources.DefaultWriterContainer.writeRows(WriterContainer.scala:261)
at org.apache.spark.sql.execution.datasources.InsertIntoHadoopFsRelationCommand$$anonfun$run$1$$anonfun$apply$mcV$sp$1.apply(InsertIntoHadoopFsRelationCommand.scala:143)
at org.apache.spark.sql.execution.datasources.InsertIntoHadoopFsRelationCommand$$anonfun$run$1$$anonfun$apply$mcV$sp$1.apply(InsertIntoHadoopFsRelationCommand.scala:143)
at org.apache.spark.scheduler.ResultTask.runTask(ResultTask.scala:70)
at org.apache.spark.scheduler.Task.run(Task.scala:86)
at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:274)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)
at java.lang.Thread.run(Thread.java:745)
Caused by: java.lang.RuntimeException: Failed to commit task
at org.apache.spark.sql.execution.datasources.DefaultWriterContainer.org$apache$spark$sql$execution$datasources$DefaultWriterContainer$$commitTask$1(WriterContainer.scala:275)
at org.apache.spark.sql.execution.datasources.DefaultWriterContainer$$anonfun$writeRows$1.apply$mcV$sp(WriterContainer.scala:257)
at org.apache.spark.sql.execution.datasources.DefaultWriterContainer$$anonfun$writeRows$1.apply(WriterContainer.scala:252)
at org.apache.spark.sql.execution.datasources.DefaultWriterContainer$$anonfun$writeRows$1.apply(WriterContainer.scala:252)
at org.apache.spark.util.Utils$.tryWithSafeFinallyAndFailureCallbacks(Utils.scala:1345)
at org.apache.spark.sql.execution.datasources.DefaultWriterContainer.writeRows(WriterContainer.scala:258)
... 8 more
Suppressed: java.lang.NullPointerException
at org.apache.parquet.hadoop.InternalParquetRecordWriter.flushRowGroupToStore(InternalParquetRecordWriter.java:147)
at org.apache.parquet.hadoop.InternalParquetRecordWriter.close(InternalParquetRecordWriter.java:113)
at org.apache.parquet.hadoop.ParquetRecordWriter.close(ParquetRecordWriter.java:112)
at org.apache.spark.sql.execution.datasources.parquet.ParquetOutputWriter.close(ParquetFileFormat.scala:569)
at org.apache.spark.sql.execution.datasources.DefaultWriterContainer.org$apache$spark$sql$execution$datasources$DefaultWriterContainer$$abortTask$1(WriterContainer.scala:282)
at org.apache.spark.sql.execution.datasources.DefaultWriterContainer$$anonfun$writeRows$2.apply$mcV$sp(WriterContainer.scala:258)
at org.apache.spark.util.Utils$.tryWithSafeFinallyAndFailureCallbacks(Utils.scala:1354)
... 9 more
Caused by: com.amazonaws.AmazonClientException: Unable to unmarshall response (Failed to parse XML document with handler class com.amazonaws.services.s3.model.transform.XmlResponsesSaxParser$ListBucketHandler). Response Code: 200, Response Text: OK
at com.amazonaws.http.AmazonHttpClient.handleResponse(AmazonHttpClient.java:738)
at com.amazonaws.http.AmazonHttpClient.executeHelper(AmazonHttpClient.java:399)
at com.amazonaws.http.AmazonHttpClient.execute(AmazonHttpClient.java:232)
at com.amazonaws.services.s3.AmazonS3Client.invoke(AmazonS3Client.java:3528)
at com.amazonaws.services.s3.AmazonS3Client.invoke(AmazonS3Client.java:3480)
at com.amazonaws.services.s3.AmazonS3Client.listObjects(AmazonS3Client.java:604)
at org.apache.hadoop.fs.s3a.S3AFileSystem.getFileStatus(S3AFileSystem.java:962)
at org.apache.hadoop.fs.s3a.S3AFileSystem.deleteUnnecessaryFakeDirectories(S3AFileSystem.java:1147)
at org.apache.hadoop.fs.s3a.S3AFileSystem.finishedWrite(S3AFileSystem.java:1136)
at org.apache.hadoop.fs.s3a.S3AOutputStream.close(S3AOutputStream.java:142)
at org.apache.hadoop.fs.FSDataOutputStream$PositionCache.close(FSDataOutputStream.java:72)
at org.apache.hadoop.fs.FSDataOutputStream.close(FSDataOutputStream.java:106)
at org.apache.parquet.hadoop.ParquetFileWriter.end(ParquetFileWriter.java:400)
at org.apache.parquet.hadoop.InternalParquetRecordWriter.close(InternalParquetRecordWriter.java:117)
at org.apache.parquet.hadoop.ParquetRecordWriter.close(ParquetRecordWriter.java:112)
at org.apache.spark.sql.execution.datasources.parquet.ParquetOutputWriter.close(ParquetFileFormat.scala:569)
at org.apache.spark.sql.execution.datasources.DefaultWriterContainer.org$apache$spark$sql$execution$datasources$DefaultWriterContainer$$commitTask$1(WriterContainer.scala:267)
... 13 more
Caused by: com.amazonaws.AmazonClientException: Failed to parse XML document with handler class com.amazonaws.services.s3.model.transform.XmlResponsesSaxParser$ListBucketHandler
at com.amazonaws.services.s3.model.transform.XmlResponsesSaxParser.parseXmlInputStream(XmlResponsesSaxParser.java:150)
at com.amazonaws.services.s3.model.transform.XmlResponsesSaxParser.parseListBucketObjectsResponse(XmlResponsesSaxParser.java:279)
at com.amazonaws.services.s3.model.transform.Unmarshallers$ListObjectsUnmarshaller.unmarshall(Unmarshallers.java:75)
at com.amazonaws.services.s3.model.transform.Unmarshallers$ListObjectsUnmarshaller.unmarshall(Unmarshallers.java:72)
at com.amazonaws.services.s3.internal.S3XmlResponseHandler.handle(S3XmlResponseHandler.java:62)
at com.amazonaws.services.s3.internal.S3XmlResponseHandler.handle(S3XmlResponseHandler.java:31)
at com.amazonaws.http.AmazonHttpClient.handleResponse(AmazonHttpClient.java:712)
... 29 more
Caused by: org.xml.sax.SAXParseException; lineNumber: 1; columnNumber: 2; XML document structures must start and end within the same entity.
at org.apache.xerces.util.ErrorHandlerWrapper.createSAXParseException(Unknown Source)
at org.apache.xerces.util.ErrorHandlerWrapper.fatalError(Unknown Source)
at org.apache.xerces.impl.XMLErrorReporter.reportError(Unknown Source)
at org.apache.xerces.impl.XMLErrorReporter.reportError(Unknown Source)
at org.apache.xerces.impl.XMLErrorReporter.reportError(Unknown Source)
at org.apache.xerces.impl.XMLScanner.reportFatalError(Unknown Source)
at org.apache.xerces.impl.XMLDocumentFragmentScannerImpl.endEntity(Unknown Source)
at org.apache.xerces.impl.XMLDocumentScannerImpl.endEntity(Unknown Source)
at org.apache.xerces.impl.XMLEntityManager.endEntity(Unknown Source)
at org.apache.xerces.impl.XMLEntityScanner.load(Unknown Source)
at org.apache.xerces.impl.XMLEntityScanner.skipChar(Unknown Source)
at org.apache.xerces.impl.XMLDocumentScannerImpl$PrologDispatcher.dispatch(Unknown Source)
at org.apache.xerces.impl.XMLDocumentFragmentScannerImpl.scanDocument(Unknown Source)
at org.apache.xerces.parsers.XML11Configuration.parse(Unknown Source)
at org.apache.xerces.parsers.XML11Configuration.parse(Unknown Source)
at org.apache.xerces.parsers.XMLParser.parse(Unknown Source)
at org.apache.xerces.parsers.AbstractSAXParser.parse(Unknown Source)
at com.amazonaws.services.s3.model.transform.XmlResponsesSaxParser.parseXmlInputStream(XmlResponsesSaxParser.java:141)
... 35 more
Here's my conf:
spark.executor.extraJavaOptions -XX:+UseG1GC -XX:MaxPermSize=1G -XX:+HeapDumpOnOutOfMemoryError
spark.executor.memory 16G
spark.executor.uri https://s3.amazonaws.com/foo/spark-2.0.1-bin-hadoop2.7.tgz
spark.hadoop.fs.s3a.impl org.apache.hadoop.fs.s3a.S3AFileSystem
spark.hadoop.fs.s3a.buffer.dir /raid0/spark
spark.hadoop.fs.s3n.buffer.dir /raid0/spark
spark.hadoop.fs.s3a.connection.timeout 500000
spark.hadoop.fs.s3n.multipart.uploads.enabled true
spark.hadoop.parquet.block.size 2147483648
spark.hadoop.parquet.enable.summary-metadata false
spark.jars.packages com.databricks:spark-avro_2.11:3.0.1
spark.local.dir /raid0/spark
spark.mesos.coarse false
spark.mesos.constraints priority:1
spark.network.timeout 600
spark.rpc.message.maxSize 500
spark.speculation false
spark.sql.parquet.mergeSchema false
spark.sql.planner.externalSort true
spark.submit.deployMode client
spark.task.cpus 1

Answer 1

I can think for three possible reasons for this problem.
JVM version. AWS SDK checks for the following ones. "1.6.0_06",
"1.6.0_13", "1.6.0_17", "1.6.0_65", "1.7.0_45". If you are using one
of them, try upgrading.
Old AWS SDK. Refer to
https://github.com/aws/aws-sdk-java/issues/460 for a workaround.
If you lots of files in the directory where you are writing these files, you might be hitting https://issues.apache.org/jira/browse/HADOOP-13164. Consider increasing the timeout to larger values.

Answer 2

A SAXParseException may indicate a badly formatted XML file. Since the job fails roughly a third of the way through consistently, this means it's probably failing in the same place every time (a file whose partition is roughly a third of the way through the partition list).
Can you paste your script? It may be possible to wrap the Spark step in a try/catch loop that will print out the file if this error occurs, which will let you easily zoom in on the problem.

Answer 3

From the logs:
Caused by: org.xml.sax.SAXParseException; lineNumber: 1; columnNumber: 2; XML document structures must start and end within the same entity.
and
Caused by: com.amazonaws.AmazonClientException: Failed to parse XML document with handler class com.amazonaws.services.s3.model.transform.XmlResponsesSaxParser$ListBucketHandler
It looks like you have a corrupted/incorrectly formatted file, and your error is actually occurring during the read portion of the task. You could confirm this by trying another operation that will force the read such as count().
If confirmed, the goal would then be to find the corrupted file. You could do this by listing the s3 files, sc.parallelize() that list, and then trying to read the files in a custom function using map().
import boto3
from pyspark.sql import Row
def scanKeys(startKey, endKey):
bucket = boto3.resource('s3').Bucket('bucketName')
for obj in bucket.objects.filter(Prefix='prefix', Marker=startKey):
if obj.key < endKey:
yield obj.key
else:
return
def testFile(s3Path):
s3obj = boto3.resource('s3').Object(bucket_name='bucketName', key=key)
body = s3obj.get()['Body']
...
logic to test file format, or use a try/except and attempt to parse it
...
if fileFormatedCorrectly == True:
return Row(status='Good', key = s3Path)
else:
return Row(status='Fail', key = s3Path)
keys = list(scanKeys(startKey, endKey))
keyListRdd = sc.parallelize(keys, 1000)
keyListRdd.map(testFile).filter(lambda x: x.asDict.get('status') == 'Fail').collect()
This will return the s3 paths for the incorrectly formatted files

Answer 4

For Googlers:
If you:
have a versioned bucket
use s3a://
see ListBucketHandler and listObjects in your error message
Quick solution:
use s3:// instead of s3a://, which will use S3 driver provided by EMR
You may see this error because s3a:// in older versions uses S3::ListObjects (v1) API instead of S3::ListObjectsV2. The former would return extra info like owner, and is not robust against large number of deletion markers. Newer versions of the s3a:// driver solved this problem, but you could always use the s3:// driver instead.
Quote:
the V1 list API experience always returns 5000 entries (as set in fs.s3a.paging.maximum
except for the final entry
if you have versioning turned on in your bucket, deleted entries retain tombstone markers with references to their versions
which will surface in the S3-side of list calls, but get stripped out from the response
so...for a very large tree, you may end up S3 having to keep a channel open while is skips of thousands to millions of deleted
objects before it can find actual ones to return.
which can time out connections.
Quote:
Introducing a new version of the ListObjects (ListObjectsV2) API that allows listing objects with a large number of delete markers.
Quote:
If there are thousands of delete markers, the list operation might timeout。