@Test
public void testEqual() throws Exception {
if (conf.getBoolean("tajo.storage.manager.v2", false)) {
return;
}
this.rndKey = rnd.nextInt(250);
final String QUERY = "select * from employee where managerId = " + rndKey;
FileFragment[] frags =
StorageManager.splitNG(conf, "default.employee", meta, tablePath, Integer.MAX_VALUE);
Path workDir = CommonTestingUtil.getTestDir("target/test-data/testEqual");
TaskAttemptContext ctx =
new TaskAttemptContext(
conf,
LocalTajoTestingUtility.newQueryUnitAttemptId(),
new FileFragment[] {frags[0]},
workDir);
Expr expr = analyzer.parse(QUERY);
LogicalPlan plan = planner.createPlan(LocalTajoTestingUtility.createDummySession(), expr);
LogicalNode rootNode = optimizer.optimize(plan);
TmpPlanner phyPlanner = new TmpPlanner(conf, sm);
PhysicalExec exec = phyPlanner.createPlan(ctx, rootNode);
int tupleCount = this.randomValues.get(rndKey);
int counter = 0;
exec.init();
while (exec.next() != null) {
counter++;
}
exec.close();
assertEquals(tupleCount, counter);
}
@Test
public void testFetchImpl() {
ExecutionBlockId ebId = new ExecutionBlockId(LocalTajoTestingUtility.newQueryId(), 0);
Task.PullHost pullHost = new Task.PullHost("localhost", 0);
FetchImpl expected = new FetchImpl(pullHost, SCATTERED_HASH_SHUFFLE, ebId, 1);
FetchImpl fetch2 = new FetchImpl(pullHost, SCATTERED_HASH_SHUFFLE, ebId, 1);
assertEquals(expected, fetch2);
fetch2.setOffset(5);
fetch2.setLength(10);
assertNotEquals(expected, fetch2);
}
@Test
public void testScheduleFetchesByEvenDistributedVolumes() {
Map<Integer, FetchGroupMeta> fetchGroups = Maps.newHashMap();
String tableName = "test1";
ExecutionBlockId ebId = new ExecutionBlockId(LocalTajoTestingUtility.newQueryId(), 0);
FetchImpl[] fetches = new FetchImpl[12];
for (int i = 0; i < 12; i++) {
fetches[i] =
new FetchImpl(new Task.PullHost("localhost", 10000 + i), HASH_SHUFFLE, ebId, i / 2);
}
int[] VOLUMES = {100, 80, 70, 30, 10, 5};
for (int i = 0; i < 12; i += 2) {
fetchGroups.put(i, new FetchGroupMeta(VOLUMES[i / 2], fetches[i]).addFetche(fetches[i + 1]));
}
Pair<Long[], Map<String, List<FetchImpl>>[]> results;
results = Repartitioner.makeEvenDistributedFetchImpl(fetchGroups, tableName, 1);
long expected[] = {100 + 80 + 70 + 30 + 10 + 5};
assertFetchVolumes(expected, results.getFirst());
assertFetchImpl(fetches, results.getSecond());
results = Repartitioner.makeEvenDistributedFetchImpl(fetchGroups, tableName, 2);
long expected0[] = {140, 155};
assertFetchVolumes(expected0, results.getFirst());
assertFetchImpl(fetches, results.getSecond());
results = Repartitioner.makeEvenDistributedFetchImpl(fetchGroups, tableName, 3);
long expected1[] = {100, 95, 100};
assertFetchVolumes(expected1, results.getFirst());
assertFetchImpl(fetches, results.getSecond());
results = Repartitioner.makeEvenDistributedFetchImpl(fetchGroups, tableName, 4);
long expected2[] = {100, 80, 70, 45};
assertFetchVolumes(expected2, results.getFirst());
assertFetchImpl(fetches, results.getSecond());
results = Repartitioner.makeEvenDistributedFetchImpl(fetchGroups, tableName, 5);
long expected3[] = {100, 80, 70, 30, 15};
assertFetchVolumes(expected3, results.getFirst());
assertFetchImpl(fetches, results.getSecond());
results = Repartitioner.makeEvenDistributedFetchImpl(fetchGroups, tableName, 6);
long expected4[] = {100, 80, 70, 30, 10, 5};
assertFetchVolumes(expected4, results.getFirst());
assertFetchImpl(fetches, results.getSecond());
}
@Test
public final void testRightOuter_MergeJoin3() throws IOException, TajoException {
Expr expr = analyzer.parse(QUERIES[3]);
LogicalNode plan = planner.createPlan(defaultContext, expr).getRootBlock().getRoot();
JoinNode joinNode = PlannerUtil.findTopNode(plan, NodeType.JOIN);
Enforcer enforcer = new Enforcer();
enforcer.enforceJoinAlgorithm(joinNode.getPID(), JoinAlgorithm.MERGE_JOIN);
FileFragment[] emp3Frags =
FileTablespace.splitNG(
conf, EMP3_NAME, emp3.getMeta(), new Path(emp3.getUri()), Integer.MAX_VALUE);
FileFragment[] dep4Frags =
FileTablespace.splitNG(
conf, DEP4_NAME, dep4.getMeta(), new Path(dep4.getUri()), Integer.MAX_VALUE);
FileFragment[] merged = TUtil.concat(emp3Frags, dep4Frags);
Path workDir =
CommonTestingUtil.getTestDir(
TajoTestingCluster.DEFAULT_TEST_DIRECTORY + "/testRightOuter_MergeJoin3");
TaskAttemptContext ctx =
new TaskAttemptContext(
new QueryContext(conf), LocalTajoTestingUtility.newTaskAttemptId(), merged, workDir);
ctx.setEnforcer(enforcer);
PhysicalPlanner phyPlanner = new PhysicalPlannerImpl(conf);
PhysicalExec exec = phyPlanner.createPlan(ctx, plan);
ProjectionExec proj = (ProjectionExec) exec;
assertTrue(proj.getChild() instanceof RightOuterMergeJoinExec);
int count = 0;
exec.init();
while (exec.next() != null) {
// TODO check contents
count = count + 1;
}
assertNull(exec.next());
exec.close();
assertEquals(13, count);
}
@Before
public void setUp() throws Exception {
util = new TajoTestingCluster();
util.initTestDir();
util.startCatalogCluster();
catalog = util.getCatalogService();
testDir = CommonTestingUtil.getTestDir(TEST_PATH);
catalog.createTablespace(DEFAULT_TABLESPACE_NAME, testDir.toUri().toString());
catalog.createDatabase(DEFAULT_DATABASE_NAME, DEFAULT_TABLESPACE_NAME);
conf = util.getConfiguration();
// ----------------- dep3 ------------------------------
// dep_id | dep_name | loc_id
// --------------------------------
// 0 | dep_0 | 1000
// 1 | dep_1 | 1001
// 2 | dep_2 | 1002
// 3 | dep_3 | 1003
// 4 | dep_4 | 1004
// 5 | dep_5 | 1005
// 6 | dep_6 | 1006
// 7 | dep_7 | 1007
// 8 | dep_8 | 1008
// 9 | dep_9 | 1009
Schema dep3Schema = new Schema();
dep3Schema.addColumn("dep_id", Type.INT4);
dep3Schema.addColumn("dep_name", Type.TEXT);
dep3Schema.addColumn("loc_id", Type.INT4);
TableMeta dep3Meta = CatalogUtil.newTableMeta("TEXT");
Path dep3Path = new Path(testDir, "dep3.csv");
Appender appender1 =
((FileTablespace) TablespaceManager.getLocalFs())
.getAppender(dep3Meta, dep3Schema, dep3Path);
appender1.init();
VTuple tuple = new VTuple(dep3Schema.size());
for (int i = 0; i < 10; i++) {
tuple.put(
new Datum[] {
DatumFactory.createInt4(i),
DatumFactory.createText("dept_" + i),
DatumFactory.createInt4(1000 + i)
});
appender1.addTuple(tuple);
}
appender1.flush();
appender1.close();
dep3 = CatalogUtil.newTableDesc(DEP3_NAME, dep3Schema, dep3Meta, dep3Path);
catalog.createTable(dep3);
// ----------------- dep4 ------------------------------
// dep_id | dep_name | loc_id
// --------------------------------
// 0 | dep_0 | 1000
// 1 | dep_1 | 1001
// 2 | dep_2 | 1002
// 3 | dep_3 | 1003
// 4 | dep_4 | 1004
// 5 | dep_5 | 1005
// 6 | dep_6 | 1006
// 7 | dep_7 | 1007
// 8 | dep_8 | 1008
// 9 | dep_9 | 1009
// 10 | dep_10 | 1010
Schema dep4Schema = new Schema();
dep4Schema.addColumn("dep_id", Type.INT4);
dep4Schema.addColumn("dep_name", Type.TEXT);
dep4Schema.addColumn("loc_id", Type.INT4);
TableMeta dep4Meta = CatalogUtil.newTableMeta("TEXT");
Path dep4Path = new Path(testDir, "dep4.csv");
Appender appender4 =
((FileTablespace) TablespaceManager.getLocalFs())
.getAppender(dep4Meta, dep4Schema, dep4Path);
appender4.init();
VTuple tuple4 = new VTuple(dep4Schema.size());
for (int i = 0; i < 11; i++) {
tuple4.put(
new Datum[] {
DatumFactory.createInt4(i),
DatumFactory.createText("dept_" + i),
DatumFactory.createInt4(1000 + i)
});
appender4.addTuple(tuple4);
}
appender4.flush();
appender4.close();
dep4 = CatalogUtil.newTableDesc(DEP4_NAME, dep4Schema, dep4Meta, dep4Path);
catalog.createTable(dep4);
// ----------------- job3 ------------------------------
// job_id | job_title
// ----------------------
// 101 | job_101
// 102 | job_102
// 103 | job_103
Schema job3Schema = new Schema();
job3Schema.addColumn("job_id", Type.INT4);
job3Schema.addColumn("job_title", Type.TEXT);
TableMeta job3Meta = CatalogUtil.newTableMeta("TEXT");
Path job3Path = new Path(testDir, "job3.csv");
Appender appender2 =
((FileTablespace) TablespaceManager.getLocalFs())
.getAppender(job3Meta, job3Schema, job3Path);
appender2.init();
VTuple tuple2 = new VTuple(job3Schema.size());
for (int i = 1; i < 4; i++) {
int x = 100 + i;
tuple2.put(
new Datum[] {DatumFactory.createInt4(100 + i), DatumFactory.createText("job_" + x)});
appender2.addTuple(tuple2);
}
appender2.flush();
appender2.close();
job3 = CatalogUtil.newTableDesc(JOB3_NAME, job3Schema, job3Meta, job3Path);
catalog.createTable(job3);
// ---------------------emp3 --------------------
// emp_id | first_name | last_name | dep_id | salary | job_id
// ------------------------------------------------------------
// 11 | fn_11 | ln_11 | 1 | 123 | 101
// 13 | fn_13 | ln_13 | 3 | 369 | 103
// 15 | fn_15 | ln_15 | 5 | 615 | null
// 17 | fn_17 | ln_17 | 7 | 861 | null
// 19 | fn_19 | ln_19 | 9 | 1107 | null
// 21 | fn_21 | ln_21 | 1 | 123 | 101
// 23 | fn_23 | ln_23 | 3 | 369 | 103
Schema emp3Schema = new Schema();
emp3Schema.addColumn("emp_id", Type.INT4);
emp3Schema.addColumn("first_name", Type.TEXT);
emp3Schema.addColumn("last_name", Type.TEXT);
emp3Schema.addColumn("dep_id", Type.INT4);
emp3Schema.addColumn("salary", Type.FLOAT4);
emp3Schema.addColumn("job_id", Type.INT4);
TableMeta emp3Meta = CatalogUtil.newTableMeta("TEXT");
Path emp3Path = new Path(testDir, "emp3.csv");
Appender appender3 =
((FileTablespace) TablespaceManager.getLocalFs())
.getAppender(emp3Meta, emp3Schema, emp3Path);
appender3.init();
VTuple tuple3 = new VTuple(emp3Schema.size());
for (int i = 1; i < 4; i += 2) {
int x = 10 + i;
tuple3.put(
new Datum[] {
DatumFactory.createInt4(10 + i),
DatumFactory.createText("firstname_" + x),
DatumFactory.createText("lastname_" + x),
DatumFactory.createInt4(i),
DatumFactory.createFloat4(123 * i),
DatumFactory.createInt4(100 + i)
});
appender3.addTuple(tuple3);
int y = 20 + i;
tuple3.put(
new Datum[] {
DatumFactory.createInt4(20 + i),
DatumFactory.createText("firstname_" + y),
DatumFactory.createText("lastname_" + y),
DatumFactory.createInt4(i),
DatumFactory.createFloat4(123 * i),
DatumFactory.createInt4(100 + i)
});
appender3.addTuple(tuple3);
}
for (int i = 5; i < 10; i += 2) {
int x = 10 + i;
tuple3.put(
new Datum[] {
DatumFactory.createInt4(10 + i),
DatumFactory.createText("firstname_" + x),
DatumFactory.createText("lastname_" + x),
DatumFactory.createInt4(i),
DatumFactory.createFloat4(123 * i),
DatumFactory.createNullDatum()
});
appender3.addTuple(tuple3);
}
appender3.flush();
appender3.close();
emp3 = CatalogUtil.newTableDesc(EMP3_NAME, emp3Schema, emp3Meta, emp3Path);
catalog.createTable(emp3);
// ---------------------phone3 --------------------
// emp_id | phone_number
// -----------------------------------------------
// this table is empty, no rows
Schema phone3Schema = new Schema();
phone3Schema.addColumn("emp_id", Type.INT4);
phone3Schema.addColumn("phone_number", Type.TEXT);
TableMeta phone3Meta = CatalogUtil.newTableMeta("TEXT");
Path phone3Path = new Path(testDir, "phone3.csv");
Appender appender5 =
((FileTablespace) TablespaceManager.getLocalFs())
.getAppender(phone3Meta, phone3Schema, phone3Path);
appender5.init();
appender5.flush();
appender5.close();
phone3 = CatalogUtil.newTableDesc(PHONE3_NAME, phone3Schema, phone3Meta, phone3Path);
catalog.createTable(phone3);
analyzer = new SQLAnalyzer();
planner = new LogicalPlanner(catalog, TablespaceManager.getInstance());
defaultContext = LocalTajoTestingUtility.createDummyContext(conf);
}
@Test
public final void testNext() throws IOException, PlanningException {
FileFragment[] frags =
StorageManager.splitNG(
conf, "employee", employee.getMeta(), employee.getPath(), Integer.MAX_VALUE);
Path workDir = new Path(testDir, TestExternalSortExec.class.getName());
TaskAttemptContext ctx =
new TaskAttemptContext(
conf,
LocalTajoTestingUtility.newQueryUnitAttemptId(),
new FileFragment[] {frags[0]},
workDir);
ctx.setEnforcer(new Enforcer());
Expr expr = analyzer.parse(QUERIES[0]);
LogicalPlan plan = planner.createPlan(expr);
LogicalNode rootNode = plan.getRootBlock().getRoot();
PhysicalPlanner phyPlanner = new PhysicalPlannerImpl(conf, sm);
PhysicalExec exec = phyPlanner.createPlan(ctx, rootNode);
ProjectionExec proj = (ProjectionExec) exec;
// TODO - should be planed with user's optimization hint
if (!(proj.getChild() instanceof ExternalSortExec)) {
UnaryPhysicalExec sortExec = proj.getChild();
SeqScanExec scan = sortExec.getChild();
ExternalSortExec extSort =
new ExternalSortExec(ctx, sm, ((MemSortExec) sortExec).getPlan(), scan);
proj.setChild(extSort);
}
Tuple tuple;
Tuple preVal = null;
Tuple curVal;
int cnt = 0;
exec.init();
long start = System.currentTimeMillis();
TupleComparator comparator =
new TupleComparator(
proj.getSchema(),
new SortSpec[] {
new SortSpec(new Column("managerId", Type.INT4)),
new SortSpec(new Column("empId", Type.INT4))
});
while ((tuple = exec.next()) != null) {
curVal = tuple;
if (preVal != null) {
assertTrue(
"prev: " + preVal + ", but cur: " + curVal, comparator.compare(preVal, curVal) <= 0);
}
preVal = curVal;
cnt++;
}
long end = System.currentTimeMillis();
assertEquals(numTuple, cnt);
// for rescan test
preVal = null;
exec.rescan();
cnt = 0;
while ((tuple = exec.next()) != null) {
curVal = tuple;
if (preVal != null) {
assertTrue(
"prev: " + preVal + ", but cur: " + curVal, comparator.compare(preVal, curVal) <= 0);
}
preVal = curVal;
cnt++;
}
assertEquals(numTuple, cnt);
exec.close();
System.out.println("Sort Time: " + (end - start) + " msc");
}