@Test
public void testHadoopBuild() throws Exception {
// create test data
Map<String, String> values = new HashMap<String, String>();
File testDir = TestUtils.createTempDir();
File tempDir = new File(testDir, "temp"), tempDir2 = new File(testDir, "temp2");
File outputDir = new File(testDir, "output"), outputDir2 = new File(testDir, "output2");
File storeDir = TestUtils.createTempDir(testDir);
for (int i = 0; i < 200; i++) values.put(Integer.toString(i), Integer.toBinaryString(i));
// write test data to text file
File inputFile = File.createTempFile("input", ".txt", testDir);
inputFile.deleteOnExit();
StringBuilder contents = new StringBuilder();
for (Map.Entry<String, String> entry : values.entrySet())
contents.append(entry.getKey() + "\t" + entry.getValue() + "\n");
FileUtils.writeStringToFile(inputFile, contents.toString());
String storeName = "test";
SerializerDefinition serDef = new SerializerDefinition("string");
Cluster cluster = ServerTestUtils.getLocalCluster(1);
// Test backwards compatibility
StoreDefinition def =
new StoreDefinitionBuilder()
.setName(storeName)
.setType(ReadOnlyStorageConfiguration.TYPE_NAME)
.setKeySerializer(serDef)
.setValueSerializer(serDef)
.setRoutingPolicy(RoutingTier.CLIENT)
.setRoutingStrategyType(RoutingStrategyType.CONSISTENT_STRATEGY)
.setReplicationFactor(1)
.setPreferredReads(1)
.setRequiredReads(1)
.setPreferredWrites(1)
.setRequiredWrites(1)
.build();
HadoopStoreBuilder builder =
new HadoopStoreBuilder(
new Configuration(),
TextStoreMapper.class,
TextInputFormat.class,
cluster,
def,
64 * 1024,
new Path(tempDir2.getAbsolutePath()),
new Path(outputDir2.getAbsolutePath()),
new Path(inputFile.getAbsolutePath()),
CheckSumType.MD5,
saveKeys,
false);
builder.build();
builder =
new HadoopStoreBuilder(
new Configuration(),
TextStoreMapper.class,
TextInputFormat.class,
cluster,
def,
64 * 1024,
new Path(tempDir.getAbsolutePath()),
new Path(outputDir.getAbsolutePath()),
new Path(inputFile.getAbsolutePath()),
CheckSumType.MD5,
saveKeys,
false);
builder.build();
// Check if checkSum is generated in outputDir
File nodeFile = new File(outputDir, "node-0");
// Check if metadata file exists
File metadataFile = new File(nodeFile, ".metadata");
Assert.assertTrue(metadataFile.exists());
ReadOnlyStorageMetadata metadata = new ReadOnlyStorageMetadata(metadataFile);
if (saveKeys)
Assert.assertEquals(
metadata.get(ReadOnlyStorageMetadata.FORMAT),
ReadOnlyStorageFormat.READONLY_V2.getCode());
else
Assert.assertEquals(
metadata.get(ReadOnlyStorageMetadata.FORMAT),
ReadOnlyStorageFormat.READONLY_V1.getCode());
Assert.assertEquals(
metadata.get(ReadOnlyStorageMetadata.CHECKSUM_TYPE), CheckSum.toString(CheckSumType.MD5));
// Check contents of checkSum file
byte[] md5 =
Hex.decodeHex(((String) metadata.get(ReadOnlyStorageMetadata.CHECKSUM)).toCharArray());
byte[] checkSumBytes = CheckSumTests.calculateCheckSum(nodeFile.listFiles(), CheckSumType.MD5);
Assert.assertEquals(0, ByteUtils.compare(checkSumBytes, md5));
// check if fetching works
HdfsFetcher fetcher = new HdfsFetcher();
// Fetch to version directory
File versionDir = new File(storeDir, "version-0");
fetcher.fetch(nodeFile.getAbsolutePath(), versionDir.getAbsolutePath());
Assert.assertTrue(versionDir.exists());
// open store
@SuppressWarnings("unchecked")
Serializer<Object> serializer =
(Serializer<Object>) new DefaultSerializerFactory().getSerializer(serDef);
ReadOnlyStorageEngine engine =
new ReadOnlyStorageEngine(
storeName,
searchStrategy,
new RoutingStrategyFactory().updateRoutingStrategy(def, cluster),
0,
storeDir,
1);
Store<Object, Object, Object> store =
SerializingStore.wrap(engine, serializer, serializer, serializer);
// check values
for (Map.Entry<String, String> entry : values.entrySet()) {
List<Versioned<Object>> found = store.get(entry.getKey(), null);
Assert.assertEquals("Incorrect number of results", 1, found.size());
Assert.assertEquals(entry.getValue(), found.get(0).getValue());
}
// also check the iterator - first key iterator...
try {
ClosableIterator<ByteArray> keyIterator = engine.keys();
if (!saveKeys) {
fail("Should have thrown an exception since this RO format does not support iterators");
}
int numElements = 0;
while (keyIterator.hasNext()) {
Assert.assertTrue(values.containsKey(serializer.toObject(keyIterator.next().get())));
numElements++;
}
Assert.assertEquals(numElements, values.size());
} catch (UnsupportedOperationException e) {
if (saveKeys) {
fail("Should not have thrown an exception since this RO format does support iterators");
}
}
// ... and entry iterator
try {
ClosableIterator<Pair<ByteArray, Versioned<byte[]>>> entryIterator = engine.entries();
if (!saveKeys) {
fail("Should have thrown an exception since this RO format does not support iterators");
}
int numElements = 0;
while (entryIterator.hasNext()) {
Pair<ByteArray, Versioned<byte[]>> entry = entryIterator.next();
Assert.assertEquals(
values.get(serializer.toObject(entry.getFirst().get())),
serializer.toObject(entry.getSecond().getValue()));
numElements++;
}
Assert.assertEquals(numElements, values.size());
} catch (UnsupportedOperationException e) {
if (saveKeys) {
fail("Should not have thrown an exception since this RO format does support iterators");
}
}
}
public String runBuildStore(Props props, String url) throws Exception {
int replicationFactor = props.getInt("build.replication.factor", 2);
int chunkSize = props.getInt("build.chunk.size", 1024 * 1024 * 1024);
Path tempDir =
new Path(
props.getString(
"build.temp.dir", "/tmp/vold-build-and-push-" + new Random().nextLong()));
URI uri = new URI(url);
Path outputDir = new Path(props.getString("build.output.dir"), uri.getHost());
Path inputPath = getInputPath();
String keySelection = props.getString("build.key.selection", null);
String valSelection = props.getString("build.value.selection", null);
CheckSumType checkSumType =
CheckSum.fromString(props.getString("checksum.type", CheckSum.toString(CheckSumType.MD5)));
boolean saveKeys = props.getBoolean("save.keys", true);
boolean reducerPerBucket = props.getBoolean("reducer.per.bucket", false);
int numChunks = props.getInt("num.chunks", -1);
if (isAvroJob) {
String recSchema = getRecordSchema();
String keySchema = getKeySchema();
String valSchema = getValueSchema();
new VoldemortStoreBuilderJob(
this.getId() + "-build-store",
props,
new VoldemortStoreBuilderConf(
replicationFactor,
chunkSize,
tempDir,
outputDir,
inputPath,
cluster,
storeDefs,
storeName,
keySelection,
valSelection,
null,
null,
checkSumType,
saveKeys,
reducerPerBucket,
numChunks,
keyField,
valueField,
recSchema,
keySchema,
valSchema),
true)
.run();
return outputDir.toString();
}
new VoldemortStoreBuilderJob(
this.getId() + "-build-store",
props,
new VoldemortStoreBuilderConf(
replicationFactor,
chunkSize,
tempDir,
outputDir,
inputPath,
cluster,
storeDefs,
storeName,
keySelection,
valSelection,
null,
null,
checkSumType,
saveKeys,
reducerPerBucket,
numChunks))
.run();
return outputDir.toString();
}
