@Test
public void testMergeJoin() throws Exception {
String query =
"a = load '/tmp/input1';"
+ "b = load '/tmp/input2';"
+ "c = join a by $0, b by $0 using 'merge';"
+ "store c into '/tmp/output1';";
PhysicalPlan pp = Util.buildPp(pigServer, query);
MRCompiler comp = new MRCompiler(pp, pc);
comp.compile();
MROperPlan mrp = comp.getMRPlan();
assertTrue(mrp.size() == 2);
MapReduceOper mrOp0 = mrp.getRoots().get(0);
assertTrue(mrOp0.mapPlan.size() == 2);
PhysicalOperator load0 = mrOp0.mapPlan.getRoots().get(0);
MergeJoinIndexer func =
(MergeJoinIndexer)
PigContext.instantiateFuncFromSpec(((POLoad) load0).getLFile().getFuncSpec());
Field lrField = MergeJoinIndexer.class.getDeclaredField("lr");
lrField.setAccessible(true);
POLocalRearrange lr = (POLocalRearrange) lrField.get(func);
List<PhysicalPlan> innerPlans = lr.getPlans();
PhysicalOperator localrearrange0 = mrOp0.mapPlan.getSuccessors(load0).get(0);
assertTrue(localrearrange0 instanceof POLocalRearrange);
assertTrue(mrOp0.reducePlan.size() == 3);
PhysicalOperator pack0 = mrOp0.reducePlan.getRoots().get(0);
assertTrue(pack0 instanceof POPackage);
PhysicalOperator foreach0 = mrOp0.reducePlan.getSuccessors(pack0).get(0);
assertTrue(foreach0 instanceof POForEach);
PhysicalOperator store0 = mrOp0.reducePlan.getSuccessors(foreach0).get(0);
assertTrue(store0 instanceof POStore);
assertTrue(innerPlans.size() == 1);
PhysicalPlan innerPlan = innerPlans.get(0);
assertTrue(innerPlan.size() == 1);
PhysicalOperator project = innerPlan.getRoots().get(0);
assertTrue(project instanceof POProject);
assertTrue(((POProject) project).getColumn() == 0);
MapReduceOper mrOp1 = mrp.getSuccessors(mrOp0).get(0);
assertTrue(mrOp1.mapPlan.size() == 3);
PhysicalOperator load1 = mrOp1.mapPlan.getRoots().get(0);
assertTrue(load1 instanceof POLoad);
PhysicalOperator mergejoin1 = mrOp1.mapPlan.getSuccessors(load1).get(0);
assertTrue(mergejoin1 instanceof POMergeJoin);
PhysicalOperator store1 = mrOp1.mapPlan.getSuccessors(mergejoin1).get(0);
assertTrue(store1 instanceof POStore);
assertTrue(mrOp1.reducePlan.isEmpty());
}
@Test
public void testMergeJoinWithIndexableLoadFunc() throws Exception {
String query =
"a = load 'input1';"
+ "b = load 'input2' using "
+ TestMergeJoin.DummyIndexableLoader.class.getName()
+ ";"
+ "c = join a by $0, b by $0 using 'merge';"
+ "store c into 'output';";
PhysicalPlan pp = Util.buildPp(pigServer, query);
MROperPlan mp = Util.buildMRPlan(pp, pc);
assertEquals(
"Checking number of MR Jobs for merge join with " + "IndexableLoadFunc:", 1, mp.size());
}
@Test
public void testReducerNumEstimationForOrderBy() throws Exception {
// Skip the test for Tez. Tez use a different mechanism.
// Equivalent test is in TestTezAutoParallelism
Assume.assumeTrue("Skip this test for TEZ", Util.isMapredExecType(cluster.getExecType()));
// use the estimation
pc.getProperties().setProperty("pig.exec.reducers.bytes.per.reducer", "100");
pc.getProperties().setProperty("pig.exec.reducers.max", "10");
String query = "a = load '/passwd';" + "b = order a by $0;" + "store b into 'output';";
PigServer ps = new PigServer(cluster.getExecType(), cluster.getProperties());
PhysicalPlan pp = Util.buildPp(ps, query);
MROperPlan mrPlan = Util.buildMRPlanWithOptimizer(pp, pc);
Configuration conf = ConfigurationUtil.toConfiguration(pc.getProperties());
JobControlCompiler jcc = new JobControlCompiler(pc, conf);
JobControl jobControl = jcc.compile(mrPlan, query);
assertEquals(2, mrPlan.size());
// first job uses a single reducer for the sampling
Util.assertParallelValues(-1, 1, -1, 1, jobControl.getWaitingJobs().get(0).getJobConf());
// Simulate the first job having run so estimation kicks in.
MapReduceOper sort = mrPlan.getLeaves().get(0);
jcc.updateMROpPlan(jobControl.getReadyJobs());
FileLocalizer.create(sort.getQuantFile(), pc);
jobControl = jcc.compile(mrPlan, query);
sort = mrPlan.getLeaves().get(0);
long reducer =
Math.min(
(long) Math.ceil(new File("test/org/apache/pig/test/data/passwd").length() / 100.0),
10);
assertEquals(reducer, sort.getRequestedParallelism());
// the second job estimates reducers
Util.assertParallelValues(
-1, -1, reducer, reducer, jobControl.getWaitingJobs().get(0).getJobConf());
// use the PARALLEL key word, it will override the estimated reducer number
query = "a = load '/passwd';" + "b = order a by $0 PARALLEL 2;" + "store b into 'output';";
pp = Util.buildPp(ps, query);
mrPlan = Util.buildMRPlanWithOptimizer(pp, pc);
assertEquals(2, mrPlan.size());
sort = mrPlan.getLeaves().get(0);
assertEquals(2, sort.getRequestedParallelism());
// the estimation won't take effect when it apply to non-dfs or the files doesn't exist, such as
// hbase
query =
"a = load 'hbase://passwd' using org.apache.pig.backend.hadoop.hbase.HBaseStorage('c:f1 c:f2');"
+ "b = order a by $0 ;"
+ "store b into 'output';";
pp = Util.buildPp(ps, query);
mrPlan = Util.buildMRPlanWithOptimizer(pp, pc);
assertEquals(2, mrPlan.size());
sort = mrPlan.getLeaves().get(0);
// the requested parallel will be -1 if users don't set any of default_parallel, paralllel
// and the estimation doesn't take effect. MR framework will finally set it to 1.
assertEquals(-1, sort.getRequestedParallelism());
// test order by with three jobs (after optimization)
query =
"a = load '/passwd';"
+ "b = foreach a generate $0, $1, $2;"
+ "c = order b by $0;"
+ "store c into 'output';";
pp = Util.buildPp(ps, query);
mrPlan = Util.buildMRPlanWithOptimizer(pp, pc);
assertEquals(3, mrPlan.size());
// Simulate the first 2 jobs having run so estimation kicks in.
sort = mrPlan.getLeaves().get(0);
FileLocalizer.create(sort.getQuantFile(), pc);
jobControl = jcc.compile(mrPlan, query);
Util.copyFromLocalToCluster(
cluster,
"test/org/apache/pig/test/data/passwd",
((POLoad) sort.mapPlan.getRoots().get(0)).getLFile().getFileName());
// First job is just foreach with projection, mapper-only job, so estimate gets ignored
Util.assertParallelValues(-1, -1, -1, 0, jobControl.getWaitingJobs().get(0).getJobConf());
jcc.updateMROpPlan(jobControl.getReadyJobs());
jobControl = jcc.compile(mrPlan, query);
jcc.updateMROpPlan(jobControl.getReadyJobs());
// Second job is a sampler, which requests and gets 1 reducer
Util.assertParallelValues(-1, 1, -1, 1, jobControl.getWaitingJobs().get(0).getJobConf());
jobControl = jcc.compile(mrPlan, query);
sort = mrPlan.getLeaves().get(0);
assertEquals(reducer, sort.getRequestedParallelism());
// Third job is the order, which uses the estimated number of reducers
Util.assertParallelValues(
-1, -1, reducer, reducer, jobControl.getWaitingJobs().get(0).getJobConf());
}