use of com.facebook.presto.common.Page in project presto by prestodb.
the class TestHashJoinOperator method testInnerJoinWithNullProbe.
@Test(dataProvider = "hashJoinTestValues")
public void testInnerJoinWithNullProbe(boolean parallelBuild, boolean probeHashEnabled, boolean buildHashEnabled) {
TaskContext taskContext = createTaskContext();
// build factory
List<Type> buildTypes = ImmutableList.of(VARCHAR);
RowPagesBuilder buildPages = rowPagesBuilder(buildHashEnabled, Ints.asList(0), buildTypes).row("a").row("b").row("c");
BuildSideSetup buildSideSetup = setupBuildSide(parallelBuild, taskContext, Ints.asList(0), buildPages, Optional.empty(), false, SINGLE_STREAM_SPILLER_FACTORY);
JoinBridgeManager<PartitionedLookupSourceFactory> lookupSourceFactory = buildSideSetup.getLookupSourceFactoryManager();
// probe factory
List<Type> probeTypes = ImmutableList.of(VARCHAR);
RowPagesBuilder probePages = rowPagesBuilder(probeHashEnabled, Ints.asList(0), probeTypes);
List<Page> probeInput = probePages.row("a").row((String) null).row((String) null).row("a").row("b").build();
OperatorFactory joinOperatorFactory = innerJoinOperatorFactory(lookupSourceFactory, probePages, PARTITIONING_SPILLER_FACTORY);
// build drivers and operators
instantiateBuildDrivers(buildSideSetup, taskContext);
buildLookupSource(buildSideSetup);
// expected
MaterializedResult expected = MaterializedResult.resultBuilder(taskContext.getSession(), concat(probeTypes, buildPages.getTypesWithoutHash())).row("a", "a").row("a", "a").row("b", "b").build();
assertOperatorEquals(joinOperatorFactory, taskContext.addPipelineContext(0, true, true, false).addDriverContext(), probeInput, expected, true, getHashChannels(probePages, buildPages));
}
use of com.facebook.presto.common.Page in project presto by prestodb.
the class TestLimitOperator method testLimitWithPageAlignment.
@Test
public void testLimitWithPageAlignment() {
List<Page> input = rowPagesBuilder(BIGINT).addSequencePage(3, 1).addSequencePage(2, 4).addSequencePage(2, 6).build();
OperatorFactory operatorFactory = new LimitOperatorFactory(0, new PlanNodeId("test"), 5);
MaterializedResult expected = resultBuilder(driverContext.getSession(), BIGINT).page(createSequencePage(ImmutableList.of(BIGINT), 3, 1)).page(createSequencePage(ImmutableList.of(BIGINT), 2, 4)).build();
OperatorAssertion.assertOperatorEquals(operatorFactory, driverContext, input, expected);
}
use of com.facebook.presto.common.Page in project presto by prestodb.
the class GroupByHashYieldAssertion method finishOperatorWithYieldingGroupByHash.
/**
* @param operatorFactory creates an Operator that should directly or indirectly contain GroupByHash
* @param getHashCapacity returns the hash table capacity for the input operator
* @param additionalMemoryInBytes the memory used in addition to the GroupByHash in the operator (e.g., aggregator)
*/
public static GroupByHashYieldResult finishOperatorWithYieldingGroupByHash(List<Page> input, Type hashKeyType, OperatorFactory operatorFactory, Function<Operator, Integer> getHashCapacity, long additionalMemoryInBytes) {
assertLessThan(additionalMemoryInBytes, 1L << 21, "additionalMemoryInBytes should be a relatively small number");
List<Page> result = new LinkedList<>();
// mock an adjustable memory pool
QueryId queryId1 = new QueryId("test_query1");
QueryId queryId2 = new QueryId("test_query2");
MemoryPool memoryPool = new MemoryPool(new MemoryPoolId("test"), new DataSize(1, GIGABYTE));
QueryContext queryContext = new QueryContext(queryId2, new DataSize(512, MEGABYTE), new DataSize(1024, MEGABYTE), new DataSize(512, MEGABYTE), new DataSize(1, GIGABYTE), memoryPool, new TestingGcMonitor(), EXECUTOR, SCHEDULED_EXECUTOR, new DataSize(512, MEGABYTE), new SpillSpaceTracker(new DataSize(512, MEGABYTE)), listJsonCodec(TaskMemoryReservationSummary.class));
DriverContext driverContext = createTaskContext(queryContext, EXECUTOR, TEST_SESSION).addPipelineContext(0, true, true, false).addDriverContext();
Operator operator = operatorFactory.createOperator(driverContext);
// run operator
int yieldCount = 0;
long expectedReservedExtraBytes = 0;
for (Page page : input) {
// unblocked
assertTrue(operator.needsInput());
// saturate the pool with a tiny memory left
long reservedMemoryInBytes = memoryPool.getFreeBytes() - additionalMemoryInBytes;
memoryPool.reserve(queryId1, "test", reservedMemoryInBytes);
long oldMemoryUsage = operator.getOperatorContext().getDriverContext().getMemoryUsage();
int oldCapacity = getHashCapacity.apply(operator);
// add a page and verify different behaviors
operator.addInput(page);
// get output to consume the input
Page output = operator.getOutput();
if (output != null) {
result.add(output);
}
long newMemoryUsage = operator.getOperatorContext().getDriverContext().getMemoryUsage();
// between rehash and memory used by aggregator
if (newMemoryUsage < new DataSize(4, MEGABYTE).toBytes()) {
// free the pool for the next iteration
memoryPool.free(queryId1, "test", reservedMemoryInBytes);
// this required in case input is blocked
operator.getOutput();
continue;
}
long actualIncreasedMemory = newMemoryUsage - oldMemoryUsage;
if (operator.needsInput()) {
// We have successfully added a page
// Assert we are not blocked
assertTrue(operator.getOperatorContext().isWaitingForMemory().isDone());
// assert the hash capacity is not changed; otherwise, we should have yielded
assertTrue(oldCapacity == getHashCapacity.apply(operator));
// We are not going to rehash; therefore, assert the memory increase only comes from the aggregator
assertLessThan(actualIncreasedMemory, additionalMemoryInBytes);
// free the pool for the next iteration
memoryPool.free(queryId1, "test", reservedMemoryInBytes);
} else {
// We failed to finish the page processing i.e. we yielded
yieldCount++;
// Assert we are blocked
assertFalse(operator.getOperatorContext().isWaitingForMemory().isDone());
// Hash table capacity should not change
assertEquals(oldCapacity, (long) getHashCapacity.apply(operator));
// Increased memory is no smaller than the hash table size and no greater than the hash table size + the memory used by aggregator
if (hashKeyType == BIGINT) {
// groupIds and values double by hashCapacity; while valuesByGroupId double by maxFill = hashCapacity / 0.75
expectedReservedExtraBytes = oldCapacity * (long) (Long.BYTES * 1.75 + Integer.BYTES) + page.getRetainedSizeInBytes();
} else {
// groupAddressByHash, groupIdsByHash, and rawHashByHashPosition double by hashCapacity; while groupAddressByGroupId double by maxFill = hashCapacity / 0.75
expectedReservedExtraBytes = oldCapacity * (long) (Long.BYTES * 1.75 + Integer.BYTES + Byte.BYTES) + page.getRetainedSizeInBytes();
}
assertBetweenInclusive(actualIncreasedMemory, expectedReservedExtraBytes, expectedReservedExtraBytes + additionalMemoryInBytes);
// Output should be blocked as well
assertNull(operator.getOutput());
// Free the pool to unblock
memoryPool.free(queryId1, "test", reservedMemoryInBytes);
// Trigger a process through getOutput() or needsInput()
output = operator.getOutput();
if (output != null) {
result.add(output);
}
assertTrue(operator.needsInput());
// Hash table capacity has increased
assertGreaterThan(getHashCapacity.apply(operator), oldCapacity);
// Assert the estimated reserved memory before rehash is very close to the one after rehash
long rehashedMemoryUsage = operator.getOperatorContext().getDriverContext().getMemoryUsage();
assertBetweenInclusive(rehashedMemoryUsage * 1.0 / newMemoryUsage, 0.99, 1.01);
// unblocked
assertTrue(operator.needsInput());
}
}
result.addAll(finishOperator(operator));
return new GroupByHashYieldResult(yieldCount, expectedReservedExtraBytes, result);
}
use of com.facebook.presto.common.Page in project presto by prestodb.
the class OperatorAssertion method finishOperator.
public static List<Page> finishOperator(Operator operator) {
ImmutableList.Builder<Page> outputPages = ImmutableList.builder();
for (int loopsSinceLastPage = 0; !operator.isFinished() && loopsSinceLastPage < 1_000; loopsSinceLastPage++) {
if (handledBlocked(operator)) {
continue;
}
operator.finish();
Page outputPage = operator.getOutput();
if (outputPage != null && outputPage.getPositionCount() != 0) {
outputPages.add(outputPage);
loopsSinceLastPage = 0;
}
// revoke memory when output pages have started being produced
handleMemoryRevoking(operator);
}
assertEquals(operator.isFinished(), true, "Operator did not finish");
assertEquals(operator.needsInput(), false, "Operator still wants input");
assertEquals(operator.isBlocked().isDone(), true, "Operator is blocked");
return outputPages.build();
}
use of com.facebook.presto.common.Page in project presto by prestodb.
the class TestAggregationOperator method testAggregation.
@Test
public void testAggregation() {
InternalAggregationFunction countVarcharColumn = getAggregation("count", VARCHAR);
InternalAggregationFunction maxVarcharColumn = getAggregation("max", VARCHAR);
List<Page> input = rowPagesBuilder(VARCHAR, BIGINT, VARCHAR, BIGINT, REAL, DOUBLE, VARCHAR).addSequencePage(100, 0, 0, 300, 500, 400, 500, 500).build();
OperatorFactory operatorFactory = new AggregationOperatorFactory(0, new PlanNodeId("test"), Step.SINGLE, ImmutableList.of(COUNT.bind(ImmutableList.of(0), Optional.empty()), LONG_SUM.bind(ImmutableList.of(1), Optional.empty()), LONG_AVERAGE.bind(ImmutableList.of(1), Optional.empty()), maxVarcharColumn.bind(ImmutableList.of(2), Optional.empty()), countVarcharColumn.bind(ImmutableList.of(0), Optional.empty()), LONG_SUM.bind(ImmutableList.of(3), Optional.empty()), REAL_SUM.bind(ImmutableList.of(4), Optional.empty()), DOUBLE_SUM.bind(ImmutableList.of(5), Optional.empty()), maxVarcharColumn.bind(ImmutableList.of(6), Optional.empty())), false);
DriverContext driverContext = createTaskContext(executor, scheduledExecutor, TEST_SESSION).addPipelineContext(0, true, true, false).addDriverContext();
MaterializedResult expected = resultBuilder(driverContext.getSession(), BIGINT, BIGINT, DOUBLE, VARCHAR, BIGINT, BIGINT, REAL, DOUBLE, VARCHAR).row(100L, 4950L, 49.5, "399", 100L, 54950L, 44950.0f, 54950.0, "599").build();
assertOperatorEquals(operatorFactory, driverContext, input, expected);
assertEquals(driverContext.getSystemMemoryUsage(), 0);
assertEquals(driverContext.getMemoryUsage(), 0);
}
Aggregations