use of org.apache.hudi.common.bootstrap.index.BootstrapIndex.IndexWriter in project hudi by apache.
the class TestHoodieTableFileSystemView method testViewForFileSlicesWithAsyncCompaction.
/**
* Helper method to test Views in the presence of concurrent compaction.
*
* @param skipCreatingDataFile if set, first File Slice will not have data-file set. This would simulate inserts going
* directly to log files
* @param isCompactionInFlight if set, compaction was inflight (running) when view was tested first time, otherwise
* compaction was in requested state
* @param expTotalFileSlices Total number of file-slices across file-groups in the partition path
* @param expTotalDataFiles Total number of data-files across file-groups in the partition path
* @param includeInvalidAndInflight Whether view includes inflight and invalid file-groups.
* @param testBootstrap enable Bootstrap and test
* @throws Exception -
*/
protected void testViewForFileSlicesWithAsyncCompaction(boolean skipCreatingDataFile, boolean isCompactionInFlight, int expTotalFileSlices, int expTotalDataFiles, boolean includeInvalidAndInflight, boolean testBootstrap) throws Exception {
if (testBootstrap) {
metaClient = HoodieTestUtils.init(tempDir.toAbsolutePath().toString(), getTableType(), BOOTSTRAP_SOURCE_PATH, testBootstrap);
}
String partitionPath = "2016/05/01";
new File(basePath + "/" + partitionPath).mkdirs();
String fileId = UUID.randomUUID().toString();
String srcName = "part_0000" + metaClient.getTableConfig().getBaseFileFormat().getFileExtension();
HoodieFileStatus srcFileStatus = HoodieFileStatus.newBuilder().setPath(HoodiePath.newBuilder().setUri(BOOTSTRAP_SOURCE_PATH + partitionPath + "/" + srcName).build()).setLength(256 * 1024 * 1024L).setAccessTime(new Date().getTime()).setModificationTime(new Date().getTime() + 99999).setBlockReplication(2).setOwner("hudi").setGroup("hudi").setBlockSize(128 * 1024 * 1024L).setPermission(HoodieFSPermission.newBuilder().setUserAction(FsAction.ALL.name()).setGroupAction(FsAction.READ.name()).setOtherAction(FsAction.NONE.name()).setStickyBit(true).build()).build();
// if skipCreatingDataFile, then instantTime1 below acts like delta-commit, otherwise it is base-commit
String instantTime1 = testBootstrap && !skipCreatingDataFile ? HoodieTimeline.METADATA_BOOTSTRAP_INSTANT_TS : "1";
String deltaInstantTime1 = "2";
String deltaInstantTime2 = "3";
String dataFileName = null;
if (!skipCreatingDataFile) {
dataFileName = FSUtils.makeDataFileName(instantTime1, TEST_WRITE_TOKEN, fileId);
new File(basePath + "/" + partitionPath + "/" + dataFileName).createNewFile();
}
String fileName1 = FSUtils.makeLogFileName(fileId, HoodieLogFile.DELTA_EXTENSION, instantTime1, 0, TEST_WRITE_TOKEN);
String fileName2 = FSUtils.makeLogFileName(fileId, HoodieLogFile.DELTA_EXTENSION, instantTime1, 1, TEST_WRITE_TOKEN);
new File(basePath + "/" + partitionPath + "/" + fileName1).createNewFile();
new File(basePath + "/" + partitionPath + "/" + fileName2).createNewFile();
HoodieActiveTimeline commitTimeline = metaClient.getActiveTimeline();
HoodieInstant instant1 = new HoodieInstant(true, HoodieTimeline.COMMIT_ACTION, instantTime1);
HoodieInstant deltaInstant2 = new HoodieInstant(true, HoodieTimeline.DELTA_COMMIT_ACTION, deltaInstantTime1);
HoodieInstant deltaInstant3 = new HoodieInstant(true, HoodieTimeline.DELTA_COMMIT_ACTION, deltaInstantTime2);
if (testBootstrap && !skipCreatingDataFile) {
try (IndexWriter writer = new HFileBootstrapIndex(metaClient).createWriter(BOOTSTRAP_SOURCE_PATH)) {
writer.begin();
BootstrapFileMapping mapping = new BootstrapFileMapping(BOOTSTRAP_SOURCE_PATH, partitionPath, partitionPath, srcFileStatus, fileId);
List<BootstrapFileMapping> b = new ArrayList<>();
b.add(mapping);
writer.appendNextPartition(partitionPath, b);
writer.finish();
}
}
saveAsComplete(commitTimeline, instant1, Option.empty());
saveAsComplete(commitTimeline, deltaInstant2, Option.empty());
saveAsComplete(commitTimeline, deltaInstant3, Option.empty());
refreshFsView();
List<FileSlice> fileSlices = rtView.getLatestFileSlices(partitionPath).collect(Collectors.toList());
assertEquals(1, fileSlices.size());
FileSlice fileSlice = fileSlices.get(0);
assertEquals(instantTime1, fileSlice.getBaseInstantTime());
if (!skipCreatingDataFile) {
assertTrue(fileSlice.getBaseFile().isPresent());
checkExternalFile(srcFileStatus, fileSlice.getBaseFile().get().getBootstrapBaseFile(), testBootstrap);
}
String compactionRequestedTime = "4";
String compactDataFileName = FSUtils.makeDataFileName(compactionRequestedTime, TEST_WRITE_TOKEN, fileId);
List<Pair<String, FileSlice>> partitionFileSlicesPairs = new ArrayList<>();
partitionFileSlicesPairs.add(Pair.of(partitionPath, fileSlices.get(0)));
HoodieCompactionPlan compactionPlan = CompactionUtils.buildFromFileSlices(partitionFileSlicesPairs, Option.empty(), Option.empty());
HoodieInstant compactionInstant;
if (isCompactionInFlight) {
// Create a Data-file but this should be skipped by view
new File(basePath + "/" + partitionPath + "/" + compactDataFileName).createNewFile();
compactionInstant = new HoodieInstant(State.INFLIGHT, HoodieTimeline.COMPACTION_ACTION, compactionRequestedTime);
HoodieInstant requested = HoodieTimeline.getCompactionRequestedInstant(compactionInstant.getTimestamp());
commitTimeline.saveToCompactionRequested(requested, TimelineMetadataUtils.serializeCompactionPlan(compactionPlan));
commitTimeline.transitionCompactionRequestedToInflight(requested);
} else {
compactionInstant = new HoodieInstant(State.REQUESTED, HoodieTimeline.COMPACTION_ACTION, compactionRequestedTime);
commitTimeline.saveToCompactionRequested(compactionInstant, TimelineMetadataUtils.serializeCompactionPlan(compactionPlan));
}
// View immediately after scheduling compaction
refreshFsView();
List<FileSlice> slices = rtView.getLatestFileSlices(partitionPath).collect(Collectors.toList());
assertEquals(1, slices.size(), "Expected latest file-slices");
assertEquals(compactionRequestedTime, slices.get(0).getBaseInstantTime(), "Base-Instant must be compaction Instant");
assertFalse(slices.get(0).getBaseFile().isPresent(), "Latest File Slice must not have data-file");
assertEquals(0, slices.get(0).getLogFiles().count(), "Latest File Slice must not have any log-files");
// Fake delta-ingestion after compaction-requested
String deltaInstantTime4 = "5";
String deltaInstantTime5 = "6";
List<String> allInstantTimes = Arrays.asList(instantTime1, deltaInstantTime1, deltaInstantTime2, compactionRequestedTime, deltaInstantTime4, deltaInstantTime5);
String fileName3 = FSUtils.makeLogFileName(fileId, HoodieLogFile.DELTA_EXTENSION, compactionRequestedTime, 0, TEST_WRITE_TOKEN);
String fileName4 = FSUtils.makeLogFileName(fileId, HoodieLogFile.DELTA_EXTENSION, compactionRequestedTime, 1, TEST_WRITE_TOKEN);
new File(basePath + "/" + partitionPath + "/" + fileName3).createNewFile();
new File(basePath + "/" + partitionPath + "/" + fileName4).createNewFile();
HoodieInstant deltaInstant4 = new HoodieInstant(true, HoodieTimeline.DELTA_COMMIT_ACTION, deltaInstantTime4);
HoodieInstant deltaInstant5 = new HoodieInstant(true, HoodieTimeline.DELTA_COMMIT_ACTION, deltaInstantTime5);
saveAsComplete(commitTimeline, deltaInstant4, Option.empty());
saveAsComplete(commitTimeline, deltaInstant5, Option.empty());
refreshFsView();
List<HoodieBaseFile> dataFiles = roView.getAllBaseFiles(partitionPath).collect(Collectors.toList());
if (skipCreatingDataFile) {
assertTrue(dataFiles.isEmpty(), "No data file expected");
} else {
assertEquals(1, dataFiles.size(), "One data-file is expected as there is only one file-group");
assertEquals(dataFileName, dataFiles.get(0).getFileName(), "Expect only valid data-file");
}
// Merge API Tests
List<FileSlice> fileSliceList = rtView.getLatestMergedFileSlicesBeforeOrOn(partitionPath, deltaInstantTime5).collect(Collectors.toList());
assertEquals(1, fileSliceList.size(), "Expect file-slice to be merged");
fileSlice = fileSliceList.get(0);
assertEquals(fileId, fileSlice.getFileId());
if (!skipCreatingDataFile) {
assertEquals(dataFileName, fileSlice.getBaseFile().get().getFileName(), "Data file must be present");
checkExternalFile(srcFileStatus, fileSlice.getBaseFile().get().getBootstrapBaseFile(), testBootstrap);
} else {
assertFalse(fileSlice.getBaseFile().isPresent(), "No data-file expected as it was not created");
}
assertEquals(instantTime1, fileSlice.getBaseInstantTime(), "Base Instant of penultimate file-slice must be base instant");
List<HoodieLogFile> logFiles = fileSlice.getLogFiles().collect(Collectors.toList());
assertEquals(4, logFiles.size(), "Log files must include those after compaction request");
assertEquals(fileName4, logFiles.get(0).getFileName(), "Log File Order check");
assertEquals(fileName3, logFiles.get(1).getFileName(), "Log File Order check");
assertEquals(fileName2, logFiles.get(2).getFileName(), "Log File Order check");
assertEquals(fileName1, logFiles.get(3).getFileName(), "Log File Order check");
fileSliceList = rtView.getLatestFileSlicesBeforeOrOn(partitionPath, deltaInstantTime5, true).collect(Collectors.toList());
assertEquals(1, fileSliceList.size(), "Expect only one file-id");
fileSlice = fileSliceList.get(0);
assertEquals(fileId, fileSlice.getFileId());
assertFalse(fileSlice.getBaseFile().isPresent(), "No data-file expected in latest file-slice");
assertEquals(compactionRequestedTime, fileSlice.getBaseInstantTime(), "Compaction requested instant must be base instant");
logFiles = fileSlice.getLogFiles().collect(Collectors.toList());
assertEquals(2, logFiles.size(), "Log files must include only those after compaction request");
assertEquals(fileName4, logFiles.get(0).getFileName(), "Log File Order check");
assertEquals(fileName3, logFiles.get(1).getFileName(), "Log File Order check");
// Data Files API tests
dataFiles = roView.getLatestBaseFiles().collect(Collectors.toList());
if (skipCreatingDataFile) {
assertEquals(0, dataFiles.size(), "Expect no data file to be returned");
} else {
assertEquals(1, dataFiles.size(), "Expect only one data-file to be sent");
dataFiles.forEach(df -> assertEquals(df.getCommitTime(), instantTime1, "Expect data-file for instant 1 be returned"));
checkExternalFile(srcFileStatus, dataFiles.get(0).getBootstrapBaseFile(), testBootstrap);
}
dataFiles = roView.getLatestBaseFiles(partitionPath).collect(Collectors.toList());
if (skipCreatingDataFile) {
assertEquals(0, dataFiles.size(), "Expect no data file to be returned");
} else {
assertEquals(1, dataFiles.size(), "Expect only one data-file to be sent");
dataFiles.forEach(df -> assertEquals(df.getCommitTime(), instantTime1, "Expect data-file for instant 1 be returned"));
checkExternalFile(srcFileStatus, dataFiles.get(0).getBootstrapBaseFile(), testBootstrap);
}
dataFiles = roView.getLatestBaseFilesBeforeOrOn(partitionPath, deltaInstantTime5).collect(Collectors.toList());
if (skipCreatingDataFile) {
assertEquals(0, dataFiles.size(), "Expect no data file to be returned");
} else {
assertEquals(1, dataFiles.size(), "Expect only one data-file to be sent");
dataFiles.forEach(df -> assertEquals(df.getCommitTime(), instantTime1, "Expect data-file for instant 1 be returned"));
checkExternalFile(srcFileStatus, dataFiles.get(0).getBootstrapBaseFile(), testBootstrap);
}
dataFiles = roView.getLatestBaseFilesInRange(allInstantTimes).collect(Collectors.toList());
if (skipCreatingDataFile) {
assertEquals(0, dataFiles.size(), "Expect no data file to be returned");
} else {
assertEquals(1, dataFiles.size(), "Expect only one data-file to be sent");
dataFiles.forEach(df -> assertEquals(df.getCommitTime(), instantTime1, "Expect data-file for instant 1 be returned"));
checkExternalFile(srcFileStatus, dataFiles.get(0).getBootstrapBaseFile(), testBootstrap);
}
// Inflight/Orphan File-groups needs to be in the view
// There is a data-file with this inflight file-id
final String inflightFileId1 = UUID.randomUUID().toString();
// There is a log-file with this inflight file-id
final String inflightFileId2 = UUID.randomUUID().toString();
// There is an orphan data file with this file-id
final String orphanFileId1 = UUID.randomUUID().toString();
// There is an orphan log data file with this file-id
final String orphanFileId2 = UUID.randomUUID().toString();
final String invalidInstantId = "INVALIDTIME";
String inflightDeltaInstantTime = "7";
String orphanDataFileName = FSUtils.makeDataFileName(invalidInstantId, TEST_WRITE_TOKEN, orphanFileId1);
new File(basePath + "/" + partitionPath + "/" + orphanDataFileName).createNewFile();
String orphanLogFileName = FSUtils.makeLogFileName(orphanFileId2, HoodieLogFile.DELTA_EXTENSION, invalidInstantId, 0, TEST_WRITE_TOKEN);
new File(basePath + "/" + partitionPath + "/" + orphanLogFileName).createNewFile();
String inflightDataFileName = FSUtils.makeDataFileName(inflightDeltaInstantTime, TEST_WRITE_TOKEN, inflightFileId1);
new File(basePath + "/" + partitionPath + "/" + inflightDataFileName).createNewFile();
String inflightLogFileName = FSUtils.makeLogFileName(inflightFileId2, HoodieLogFile.DELTA_EXTENSION, inflightDeltaInstantTime, 0, TEST_WRITE_TOKEN);
new File(basePath + "/" + partitionPath + "/" + inflightLogFileName).createNewFile();
// Mark instant as inflight
commitTimeline.createNewInstant(new HoodieInstant(State.REQUESTED, HoodieTimeline.DELTA_COMMIT_ACTION, inflightDeltaInstantTime));
commitTimeline.transitionRequestedToInflight(new HoodieInstant(State.REQUESTED, HoodieTimeline.DELTA_COMMIT_ACTION, inflightDeltaInstantTime), Option.empty());
refreshFsView();
List<FileSlice> allRawFileSlices = getAllRawFileSlices(partitionPath).collect(Collectors.toList());
dataFiles = allRawFileSlices.stream().flatMap(slice -> {
if (slice.getBaseFile().isPresent()) {
return Stream.of(slice.getBaseFile().get());
}
return Stream.empty();
}).collect(Collectors.toList());
if (includeInvalidAndInflight) {
assertEquals(2 + (isCompactionInFlight ? 1 : 0) + (skipCreatingDataFile ? 0 : 1), dataFiles.size(), "Inflight/Orphan data-file is also expected");
Set<String> fileNames = dataFiles.stream().map(HoodieBaseFile::getFileName).collect(Collectors.toSet());
assertTrue(fileNames.contains(orphanDataFileName), "Expect orphan data-file to be present");
assertTrue(fileNames.contains(inflightDataFileName), "Expect inflight data-file to be present");
if (!skipCreatingDataFile) {
assertTrue(fileNames.contains(dataFileName), "Expect old committed data-file");
}
if (isCompactionInFlight) {
assertTrue(fileNames.contains(compactDataFileName), "Expect inflight compacted data file to be present");
}
fileSliceList = getLatestRawFileSlices(partitionPath).collect(Collectors.toList());
assertEquals(includeInvalidAndInflight ? 5 : 1, fileSliceList.size(), "Expect both inflight and orphan file-slice to be included");
Map<String, FileSlice> fileSliceMap = fileSliceList.stream().collect(Collectors.toMap(FileSlice::getFileId, r -> r));
FileSlice orphanFileSliceWithDataFile = fileSliceMap.get(orphanFileId1);
FileSlice orphanFileSliceWithLogFile = fileSliceMap.get(orphanFileId2);
FileSlice inflightFileSliceWithDataFile = fileSliceMap.get(inflightFileId1);
FileSlice inflightFileSliceWithLogFile = fileSliceMap.get(inflightFileId2);
assertEquals(invalidInstantId, orphanFileSliceWithDataFile.getBaseInstantTime(), "Orphan File Slice with data-file check base-commit");
assertEquals(orphanDataFileName, orphanFileSliceWithDataFile.getBaseFile().get().getFileName(), "Orphan File Slice with data-file check data-file");
assertEquals(0, orphanFileSliceWithDataFile.getLogFiles().count(), "Orphan File Slice with data-file check data-file");
assertEquals(inflightDeltaInstantTime, inflightFileSliceWithDataFile.getBaseInstantTime(), "Inflight File Slice with data-file check base-commit");
assertEquals(inflightDataFileName, inflightFileSliceWithDataFile.getBaseFile().get().getFileName(), "Inflight File Slice with data-file check data-file");
assertEquals(0, inflightFileSliceWithDataFile.getLogFiles().count(), "Inflight File Slice with data-file check data-file");
assertEquals(invalidInstantId, orphanFileSliceWithLogFile.getBaseInstantTime(), "Orphan File Slice with log-file check base-commit");
assertFalse(orphanFileSliceWithLogFile.getBaseFile().isPresent(), "Orphan File Slice with log-file check data-file");
logFiles = orphanFileSliceWithLogFile.getLogFiles().collect(Collectors.toList());
assertEquals(1, logFiles.size(), "Orphan File Slice with log-file check data-file");
assertEquals(orphanLogFileName, logFiles.get(0).getFileName(), "Orphan File Slice with log-file check data-file");
assertEquals(inflightDeltaInstantTime, inflightFileSliceWithLogFile.getBaseInstantTime(), "Inflight File Slice with log-file check base-commit");
assertFalse(inflightFileSliceWithLogFile.getBaseFile().isPresent(), "Inflight File Slice with log-file check data-file");
logFiles = inflightFileSliceWithLogFile.getLogFiles().collect(Collectors.toList());
assertEquals(1, logFiles.size(), "Inflight File Slice with log-file check data-file");
assertEquals(inflightLogFileName, logFiles.get(0).getFileName(), "Inflight File Slice with log-file check data-file");
}
compactionInstant = new HoodieInstant(State.INFLIGHT, HoodieTimeline.COMPACTION_ACTION, compactionRequestedTime);
// Now simulate Compaction completing - Check the view
if (!isCompactionInFlight) {
// For inflight compaction, we already create a data-file to test concurrent inflight case.
// If we skipped creating data file corresponding to compaction commit, create it now
new File(basePath + "/" + partitionPath + "/" + compactDataFileName).createNewFile();
commitTimeline.createNewInstant(compactionInstant);
}
commitTimeline.saveAsComplete(compactionInstant, Option.empty());
refreshFsView();
// populate the cache
roView.getAllBaseFiles(partitionPath);
fileSliceList = rtView.getLatestFileSlices(partitionPath).collect(Collectors.toList());
LOG.info("FILESLICE LIST=" + fileSliceList);
dataFiles = fileSliceList.stream().map(FileSlice::getBaseFile).filter(Option::isPresent).map(Option::get).collect(Collectors.toList());
assertEquals(1, dataFiles.size(), "Expect only one data-files in latest view as there is only one file-group");
assertEquals(compactDataFileName, dataFiles.get(0).getFileName(), "Data Filename must match");
assertEquals(1, fileSliceList.size(), "Only one latest file-slice in the partition");
assertFalse(dataFiles.get(0).getBootstrapBaseFile().isPresent(), "No external data file must be present");
fileSlice = fileSliceList.get(0);
assertEquals(fileId, fileSlice.getFileId(), "Check file-Id is set correctly");
assertEquals(compactDataFileName, fileSlice.getBaseFile().get().getFileName(), "Check data-filename is set correctly");
assertEquals(compactionRequestedTime, fileSlice.getBaseInstantTime(), "Ensure base-instant is now compaction request instant");
logFiles = fileSlice.getLogFiles().collect(Collectors.toList());
assertEquals(2, logFiles.size(), "Only log-files after compaction request shows up");
assertEquals(fileName4, logFiles.get(0).getFileName(), "Log File Order check");
assertEquals(fileName3, logFiles.get(1).getFileName(), "Log File Order check");
// Data Files API tests
dataFiles = roView.getLatestBaseFiles().collect(Collectors.toList());
assertEquals(1, dataFiles.size(), "Expect only one data-file to be sent");
assertFalse(dataFiles.get(0).getBootstrapBaseFile().isPresent(), "No external data file must be present");
dataFiles.forEach(df -> {
assertEquals(df.getCommitTime(), compactionRequestedTime, "Expect data-file created by compaction be returned");
assertFalse(df.getBootstrapBaseFile().isPresent(), "No external data file must be present");
});
dataFiles = roView.getLatestBaseFiles(partitionPath).collect(Collectors.toList());
assertEquals(1, dataFiles.size(), "Expect only one data-file to be sent");
dataFiles.forEach(df -> {
assertEquals(df.getCommitTime(), compactionRequestedTime, "Expect data-file created by compaction be returned");
assertFalse(df.getBootstrapBaseFile().isPresent(), "No external data file must be present");
});
dataFiles = roView.getLatestBaseFilesBeforeOrOn(partitionPath, deltaInstantTime5).collect(Collectors.toList());
assertEquals(1, dataFiles.size(), "Expect only one data-file to be sent");
dataFiles.forEach(df -> {
assertEquals(df.getCommitTime(), compactionRequestedTime, "Expect data-file created by compaction be returned");
assertFalse(df.getBootstrapBaseFile().isPresent(), "No external data file must be present");
});
dataFiles = roView.getLatestBaseFilesInRange(allInstantTimes).collect(Collectors.toList());
assertEquals(1, dataFiles.size(), "Expect only one data-file to be sent");
dataFiles.forEach(df -> {
assertEquals(df.getCommitTime(), compactionRequestedTime, "Expect data-file created by compaction be returned");
assertFalse(df.getBootstrapBaseFile().isPresent(), "No external data file must be present");
});
assertEquals(expTotalFileSlices, rtView.getAllFileSlices(partitionPath).count(), "Total number of file-slices in partitions matches expected");
assertEquals(expTotalDataFiles, roView.getAllBaseFiles(partitionPath).count(), "Total number of data-files in partitions matches expected");
// file-groups includes inflight/invalid file-ids
assertEquals(5, fsView.getAllFileGroups(partitionPath).count(), "Total number of file-groups in partitions matches expected");
}
use of org.apache.hudi.common.bootstrap.index.BootstrapIndex.IndexWriter in project hudi by apache.
the class TestBootstrapIndex method generateBootstrapIndex.
public static Map<String, List<BootstrapFileMapping>> generateBootstrapIndex(HoodieTableMetaClient metaClient, String sourceBasePath, String[] partitions, int numEntriesPerPartition) {
Map<String, List<BootstrapFileMapping>> bootstrapMapping = generateBootstrapMapping(sourceBasePath, partitions, numEntriesPerPartition);
BootstrapIndex index = new HFileBootstrapIndex(metaClient);
try (IndexWriter writer = index.createWriter(sourceBasePath)) {
writer.begin();
bootstrapMapping.entrySet().stream().forEach(e -> writer.appendNextPartition(e.getKey(), e.getValue()));
writer.finish();
}
return bootstrapMapping;
}
Aggregations