1
0

[HUDI-1529] Add block size to the FileStatus objects returned from metadata table to avoid too many file splits (#2451)

This commit is contained in:
Udit Mehrotra
2021-01-18 07:29:53 -08:00
committed by GitHub
parent 3d1d5d00b0
commit 684e12e9fc
3 changed files with 16 additions and 4 deletions

View File

@@ -801,6 +801,14 @@ public class TestHoodieBackedMetadata extends HoodieClientTestHarness {
// File sizes should be valid
Arrays.stream(metaStatuses).forEach(s -> assertTrue(s.getLen() > 0));
// Block sizes should be valid
Arrays.stream(metaStatuses).forEach(s -> assertTrue(s.getBlockSize() > 0));
List<Long> fsBlockSizes = Arrays.stream(fsStatuses).map(FileStatus::getBlockSize).collect(Collectors.toList());
Collections.sort(fsBlockSizes);
List<Long> metadataBlockSizes = Arrays.stream(metaStatuses).map(FileStatus::getBlockSize).collect(Collectors.toList());
Collections.sort(metadataBlockSizes);
assertEquals(fsBlockSizes, metadataBlockSizes);
if ((fsFileNames.size() != metadataFilenames.size()) || (!fsFileNames.equals(metadataFilenames))) {
LOG.info("*** File system listing = " + Arrays.toString(fsFileNames.toArray()));
LOG.info("*** Metadata listing = " + Arrays.toString(metadataFilenames.toArray()));