use of org.apache.hudi.exception.HoodieIOException in project hudi by apache.
the class HoodieRowDataCreateHandle method makeNewPath.
private Path makeNewPath(String partitionPath) {
Path path = FSUtils.getPartitionPath(writeConfig.getBasePath(), partitionPath);
try {
if (!fs.exists(path)) {
// create a new partition as needed.
fs.mkdirs(path);
}
} catch (IOException e) {
throw new HoodieIOException("Failed to make dir " + path, e);
}
HoodieTableConfig tableConfig = table.getMetaClient().getTableConfig();
return new Path(path.toString(), FSUtils.makeDataFileName(instantTime, getWriteToken(), fileId, tableConfig.getBaseFileFormat().getFileExtension()));
}
use of org.apache.hudi.exception.HoodieIOException in project hudi by apache.
the class TestHoodieSparkMergeOnReadTableRollback method testInsertsGeneratedIntoLogFilesRollback.
@ParameterizedTest
@ValueSource(booleans = { true, false })
void testInsertsGeneratedIntoLogFilesRollback(boolean rollbackUsingMarkers) throws Exception {
Properties properties = new Properties();
properties.setProperty(HoodieTableConfig.BASE_FILE_FORMAT.key(), HoodieTableConfig.BASE_FILE_FORMAT.defaultValue().toString());
HoodieTableMetaClient metaClient = getHoodieMetaClient(HoodieTableType.MERGE_ON_READ, properties);
HoodieTestDataGenerator dataGen = new HoodieTestDataGenerator();
// insert 100 records
// Setting IndexType to be InMemory to simulate Global Index nature
HoodieWriteConfig config = getConfigBuilder(false, rollbackUsingMarkers, HoodieIndex.IndexType.INMEMORY).build();
try (SparkRDDWriteClient writeClient = getHoodieWriteClient(config)) {
String newCommitTime = "100";
writeClient.startCommitWithTime(newCommitTime);
List<HoodieRecord> records = dataGen.generateInserts(newCommitTime, 100);
JavaRDD<HoodieRecord> recordsRDD = jsc().parallelize(records, 1);
// trigger an action
List<WriteStatus> writeStatuses = ((JavaRDD<WriteStatus>) writeClient.insert(recordsRDD, newCommitTime)).collect();
// Ensure that inserts are written to only log files
assertEquals(0, writeStatuses.stream().filter(writeStatus -> !writeStatus.getStat().getPath().contains("log")).count());
assertTrue(writeStatuses.stream().anyMatch(writeStatus -> writeStatus.getStat().getPath().contains("log")));
// rollback a failed commit
boolean rollback = writeClient.rollback(newCommitTime);
assertTrue(rollback);
// insert 100 records
newCommitTime = "101";
writeClient.startCommitWithTime(newCommitTime);
records = dataGen.generateInserts(newCommitTime, 100);
recordsRDD = jsc().parallelize(records, 1);
writeClient.insert(recordsRDD, newCommitTime).collect();
// Sleep for small interval (at least 1 second) to force a new rollback start time.
Thread.sleep(1000);
// We will test HUDI-204 here. We will simulate rollback happening twice by copying the commit file to local fs
// and calling rollback twice
final String lastCommitTime = newCommitTime;
// Save the .commit file to local directory.
// Rollback will be called twice to test the case where rollback failed first time and retried.
// We got the "BaseCommitTime cannot be null" exception before the fix
java.nio.file.Path tempFolder = Files.createTempDirectory(this.getClass().getCanonicalName());
Map<String, String> fileNameMap = new HashMap<>();
for (HoodieInstant.State state : Arrays.asList(HoodieInstant.State.REQUESTED, HoodieInstant.State.INFLIGHT)) {
HoodieInstant toCopy = new HoodieInstant(state, HoodieTimeline.DELTA_COMMIT_ACTION, lastCommitTime);
File file = Files.createTempFile(tempFolder, null, null).toFile();
metaClient.getFs().copyToLocalFile(new Path(metaClient.getMetaPath(), toCopy.getFileName()), new Path(file.getAbsolutePath()));
fileNameMap.put(file.getAbsolutePath(), toCopy.getFileName());
}
Path markerDir = new Path(Files.createTempDirectory(tempFolder, null).toAbsolutePath().toString());
if (rollbackUsingMarkers) {
metaClient.getFs().copyToLocalFile(new Path(metaClient.getMarkerFolderPath(lastCommitTime)), markerDir);
}
writeClient.rollback(newCommitTime);
metaClient = HoodieTableMetaClient.reload(metaClient);
HoodieTable table = HoodieSparkTable.create(config, context());
TableFileSystemView.SliceView tableRTFileSystemView = table.getSliceView();
long numLogFiles = 0;
for (String partitionPath : dataGen.getPartitionPaths()) {
assertTrue(tableRTFileSystemView.getLatestFileSlices(partitionPath).noneMatch(fileSlice -> fileSlice.getBaseFile().isPresent()));
assertTrue(tableRTFileSystemView.getLatestFileSlices(partitionPath).noneMatch(fileSlice -> fileSlice.getLogFiles().count() > 0));
numLogFiles += tableRTFileSystemView.getLatestFileSlices(partitionPath).filter(fileSlice -> fileSlice.getLogFiles().count() > 0).count();
}
assertEquals(0, numLogFiles);
for (Map.Entry<String, String> entry : fileNameMap.entrySet()) {
try {
metaClient.getFs().copyFromLocalFile(new Path(entry.getKey()), new Path(metaClient.getMetaPath(), entry.getValue()));
} catch (IOException e) {
throw new HoodieIOException("Error copying state from local disk.", e);
}
}
if (rollbackUsingMarkers) {
metaClient.getFs().copyFromLocalFile(new Path(markerDir, lastCommitTime), new Path(metaClient.getMarkerFolderPath(lastCommitTime)));
}
Thread.sleep(1000);
// Rollback again to pretend the first rollback failed partially. This should not error out
writeClient.rollback(newCommitTime);
}
}
use of org.apache.hudi.exception.HoodieIOException in project hudi by apache.
the class HoodieAvroUtils method indexedRecordToBytes.
public static <T extends IndexedRecord> byte[] indexedRecordToBytes(T record) {
GenericDatumWriter<T> writer = new GenericDatumWriter<>(record.getSchema());
try (ByteArrayOutputStream out = new ByteArrayOutputStream()) {
BinaryEncoder encoder = EncoderFactory.get().binaryEncoder(out, reuseEncoder.get());
reuseEncoder.set(encoder);
writer.write(record, encoder);
encoder.flush();
return out.toByteArray();
} catch (IOException e) {
throw new HoodieIOException("Cannot convert GenericRecord to bytes", e);
}
}
use of org.apache.hudi.exception.HoodieIOException in project hudi by apache.
the class HFileBootstrapIndex method dropIndex.
@Override
public void dropIndex() {
try {
Path[] indexPaths = new Path[] { partitionIndexPath(metaClient), fileIdIndexPath(metaClient) };
for (Path indexPath : indexPaths) {
if (metaClient.getFs().exists(indexPath)) {
LOG.info("Dropping bootstrap index. Deleting file : " + indexPath);
metaClient.getFs().delete(indexPath);
}
}
} catch (IOException ioe) {
throw new HoodieIOException(ioe.getMessage(), ioe);
}
}
use of org.apache.hudi.exception.HoodieIOException in project hudi by apache.
the class HFileBootstrapIndex method createReader.
/**
* Helper method to create HFile Reader.
*
* @param hFilePath File Path
* @param conf Configuration
* @param fileSystem File System
*/
private static HFile.Reader createReader(String hFilePath, Configuration conf, FileSystem fileSystem) {
try {
LOG.info("Opening HFile for reading :" + hFilePath);
HFile.Reader reader = HFile.createReader(fileSystem, new HFilePathForReader(hFilePath), new CacheConfig(conf), conf);
return reader;
} catch (IOException ioe) {
throw new HoodieIOException(ioe.getMessage(), ioe);
}
}
Aggregations