Max Xie created SPARK-27267:
--------------------------------
Summary: spark 2.4 use 1.1.7.x snappy-java, but its behavior is
different from 1.1.2.x
Key: SPARK-27267
URL: https://issues.apache.org/jira/browse/SPARK-27267
Project: Spark
Issue Type: Bug
Components: Block Manager, Spark Core
Affects Versions: 2.4.0
Environment: spark.rdd.compress=true
spark.io.compression.codec =snappy
spark 2.4 in hadoop 2.6 with hive
Reporter: Max Xie
I use pyspark like that
```
from pyspark.storagelevel import StorageLevel
df=spark.sql("select * from xzn.person")
df.persist(StorageLevel(False, True, False, False))
df.count()
```
table person is a simple table stored as orc files and some orc files is empty.
When I run the query, it throw the error :
```
19/03/22 21:46:31 INFO MemoryStore:54 - Block rdd_2_1 stored as values in
memory (estimated size 0.0 B, free 1662.6 MB)
19/03/22 21:46:31 INFO FileScanRDD:54 - Reading File path:
viewfs://name/xzn.db/person/part-00011, range: 0-49, partition values: [empty
row]
19/03/22 21:46:31 INFO FileScanRDD:54 - Reading File path:
viewfs://name/xzn.db/person/part-00011_copy_1, range: 0-49, partition values:
[empty row]
19/03/22 21:46:31 INFO FileScanRDD:54 - Reading File path:
viewfs://name/xzn.db/person/part-00012, range: 0-49, partition values: [empty
row]
19/03/22 21:46:31 INFO FileScanRDD:54 - Reading File path:
viewfs://name/xzn.db/person/part-00012_copy_1, range: 0-49, partition values:
[empty row]
19/03/22 21:46:31 INFO FileScanRDD:54 - Reading File path:
viewfs://name/xzn.db/person/part-00013, range: 0-49, partition values: [empty
row]
19/03/22 21:46:31 ERROR Executor:91 - Exception in task 1.0 in stage 0.0 (TID 1)
org.xerial.snappy.SnappyIOException: [EMPTY_INPUT] Cannot decompress empty
stream
at org.xerial.snappy.SnappyInputStream.readHeader(SnappyInputStream.java:94)
at org.xerial.snappy.SnappyInputStream.<init>(SnappyInputStream.java:59)
at
org.apache.spark.io.SnappyCompressionCodec.compressedInputStream(CompressionCodec.scala:164)
at
org.apache.spark.serializer.SerializerManager.wrapForCompression(SerializerManager.scala:163)
at
org.apache.spark.serializer.SerializerManager.dataDeserializeStream(SerializerManager.scala:209)
at org.apache.spark.storage.BlockManager.getLocalValues(BlockManager.scala:596)
at
org.apache.spark.storage.BlockManager.getOrElseUpdate(BlockManager.scala:886)
at org.apache.spark.rdd.RDD.getOrCompute(RDD.scala:335)
at org.apache.spark.rdd.RDD.iterator(RDD.scala:286)
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.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.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.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.ShuffleMapTask.runTask(ShuffleMapTask.scala:99)
at org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:55)
at org.apache.spark.scheduler.Task.run(Task.scala:121)
at
org.apache.spark.executor.Executor$TaskRunner$$anonfun$10.apply(Executor.scala:402)
at org.apache.spark.util.Utils$.tryWithSafeFinally(Utils.scala:1360)
at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:408)
at
java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
at
java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
at java.lang.Thread.run(Thread.java:748)
```
After I search it, I find that 1.1.7.x snappy-java 's behavior is different
from 1.1.2.x (that spark 2.0.2 use this version). SnappyOutputStream in
1.1.2.x version always writes a snappy header whether or not to write a value,
but SnappyOutputStream in 1.1.7.x don't generate header if u don't write value
into it, so in spark 2.4 if RDD cache a empty value, memoryStore will not cache
any bytes ( no snappy header ), then it will throw the empty error.
Maybe we can change SnappyOutputStream to fix it in 1.1.7.x snappy-java, there
is my SnappyOutputStream method compressInput code
```
protected void compressInput()
throws IOException
{
// generate header
if (!headerWritten) {
outputCursor = writeHeader();
headerWritten = true;
}
if (inputCursor <= 0) {
return; // no need to dump
}
// if (!headerWritten) {
// outputCursor = writeHeader();
// headerWritten = true;
// }
// Compress and dump the buffer content
if (!hasSufficientOutputBufferFor(inputCursor)) {
dumpOutput();
}
writeBlockPreemble();
int compressedSize = Snappy.compress(inputBuffer, 0, inputCursor,
outputBuffer, outputCursor + 4);
// Write compressed data size
writeInt(outputBuffer, outputCursor, compressedSize);
outputCursor += 4 + compressedSize;
inputCursor = 0;
}
```
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
