Skip to content
This repository was archived by the owner on Nov 11, 2022. It is now read-only.

Fix HadoopFileSource’s split size estimate #534

Merged
merged 2 commits into from
Jan 26, 2017
Merged

Fix HadoopFileSource’s split size estimate #534

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
02cb720
Fix HadoopFileSource’s split size estimate
igorbernstein2 Jan 13, 2017
73a1ea6
Properly set interrupted state
igorbernstein2 Jan 26, 2017
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
11 changes: 10 additions & 1 deletion contrib/hadoop/src/main/java/com/google/cloud/dataflow/contrib/hadoop/HadoopFileSource.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -239,12 +239,21 @@ private <T> Coder<T> getDefaultCoder(Class<T> c) {
public long getEstimatedSizeBytes(PipelineOptions options) {
long size = 0;
try {
// If this source represents a split from splitIntoBundles, then return the size of the split,
// rather then the entire input
if (serializableSplit != null) {
return serializableSplit.getSplit().getLength();
}

Job job = Job.getInstance(); // new instance
for (FileStatus st : listStatus(createFormat(job), job)) {
size += st.getLen();
}
} catch (IOException | NoSuchMethodException | InvocationTargetException
| IllegalAccessException | InstantiationException e) {
| IllegalAccessException | InstantiationException) {
// ignore, and return 0
} catch (InterruptedException e) {
Thread.currentThread().interrupt();
// ignore, and return 0
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This change does not look right. At the very least, you should restore the thread interrupted state when catching an InterruptedException.

Maybe move that to its own catch block and handle specifically?

}
return size;
Expand Down
23 changes: 23 additions & 0 deletions ...b/hadoop/src/test/java/com/google/cloud/dataflow/contrib/hadoop/HadoopFileSourceTest.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -152,6 +152,29 @@ public void testSplits() throws Exception {
assertTrue(nonEmptySplits > 2);
}

@Test
public void testSplitEstimatedSize() throws Exception {
PipelineOptions options = PipelineOptionsFactory.create();

List<KV<IntWritable, Text>> expectedResults = createRandomRecords(3, 10000, 0);
File file = createFileWithData("tmp.avro", expectedResults);

HadoopFileSource<IntWritable, Text> source = HadoopFileSource.from(
file.toString(), SequenceFileInputFormat.class, IntWritable.class, Text.class
);

long originalSize = source.getEstimatedSizeBytes(options);
long splitTotalSize = 0;
List<? extends BoundedSource<KV<IntWritable, Text>>> splits = source.splitIntoBundles(
SequenceFile.SYNC_INTERVAL, options
);
for (BoundedSource<KV<IntWritable, Text>> splitSource : splits) {
splitTotalSize += splitSource.getEstimatedSizeBytes(options);
}
// Assert that the estimated size of the whole is the sum of its parts
assertEquals(originalSize, splitTotalSize);
}

private File createFileWithData(String filename, List<KV<IntWritable, Text>> records)
throws IOException {
File tmpFile = tmpFolder.newFile(filename);
Expand Down