HBASE-27666 Allow preCompact hooks to return scanners whose cells can be shipped (#5056)

Signed-off-by: Duo Zhang <zhangduo@apache.org>
This commit is contained in:
Bryan Beaudreault 2023-02-28 14:02:01 -05:00 committed by Bryan Beaudreault
parent 16b9bb8557
commit 21e3fab299
4 changed files with 162 additions and 5 deletions

View File

@ -53,6 +53,7 @@ import org.apache.hadoop.hbase.regionserver.Region.Operation;
import org.apache.hadoop.hbase.regionserver.RegionScanner;
import org.apache.hadoop.hbase.regionserver.ScanOptions;
import org.apache.hadoop.hbase.regionserver.ScanType;
import org.apache.hadoop.hbase.regionserver.Shipper;
import org.apache.hadoop.hbase.regionserver.Store;
import org.apache.hadoop.hbase.regionserver.StoreFile;
import org.apache.hadoop.hbase.regionserver.StoreFileReader;
@ -277,6 +278,15 @@ public interface RegionObserver {
* {@link InternalScanner} with a custom implementation that is returned from this method. The
* custom scanner can then inspect {@link org.apache.hadoop.hbase.Cell}s from the wrapped scanner,
* applying its own policy to what gets written.
* <p>
* If implementations are wrapping the passed in {@link InternalScanner}, they can also have their
* implementation implement {@link Shipper} and delegate to the original scanner. This will cause
* compactions to free up memory as they progress, which is especially important for people using
* off-heap memory pools.
* <p>
* Keep in mind that when {@link Shipper#shipped()} is called, any cell references you maintain in
* your implementation may get corrupted. As such you should make sure to deep clone any cells
* that you need to keep reference to across invocations of shipped.
* @param c the environment provided by the region server
* @param store the store being compacted
* @param scanner the scanner over existing data used in the store file rewriting

View File

@ -18,6 +18,7 @@
package org.apache.hadoop.hbase.regionserver;
import java.io.IOException;
import org.apache.hadoop.hbase.HBaseInterfaceAudience;
import org.apache.yetus.audience.InterfaceAudience;
/**
@ -25,7 +26,7 @@ import org.apache.yetus.audience.InterfaceAudience;
* to server and fetch N rows/RPC. These are then shipped to client. At the end of every such batch
* {@link #shipped()} will get called.
*/
@InterfaceAudience.Private
@InterfaceAudience.LimitedPrivate(HBaseInterfaceAudience.COPROC)
public interface Shipper {
/**

View File

@ -45,10 +45,10 @@ import org.apache.hadoop.hbase.regionserver.CreateStoreFileWriterParams;
import org.apache.hadoop.hbase.regionserver.HStore;
import org.apache.hadoop.hbase.regionserver.HStoreFile;
import org.apache.hadoop.hbase.regionserver.InternalScanner;
import org.apache.hadoop.hbase.regionserver.KeyValueScanner;
import org.apache.hadoop.hbase.regionserver.ScanInfo;
import org.apache.hadoop.hbase.regionserver.ScanType;
import org.apache.hadoop.hbase.regionserver.ScannerContext;
import org.apache.hadoop.hbase.regionserver.Shipper;
import org.apache.hadoop.hbase.regionserver.ShipperListener;
import org.apache.hadoop.hbase.regionserver.StoreFileReader;
import org.apache.hadoop.hbase.regionserver.StoreFileScanner;
@ -432,7 +432,7 @@ public abstract class Compactor<T extends CellSink> {
ScannerContext.newBuilder().setBatchLimit(compactionKVMax).build();
throughputController.start(compactionName);
KeyValueScanner kvs = (scanner instanceof KeyValueScanner) ? (KeyValueScanner) scanner : null;
Shipper shipper = (scanner instanceof Shipper) ? (Shipper) scanner : null;
long shippedCallSizeLimit =
(long) request.getFiles().size() * this.store.getColumnFamilyDescriptor().getBlocksize();
try {
@ -472,7 +472,7 @@ public abstract class Compactor<T extends CellSink> {
return false;
}
}
if (kvs != null && bytesWrittenProgressForShippedCall > shippedCallSizeLimit) {
if (shipper != null && bytesWrittenProgressForShippedCall > shippedCallSizeLimit) {
if (lastCleanCell != null) {
// HBASE-16931, set back sequence id to avoid affecting scan order unexpectedly.
// ShipperListener will do a clone of the last cells it refer, so need to set back
@ -488,7 +488,7 @@ public abstract class Compactor<T extends CellSink> {
// we are doing the similar thing. In between the compaction (after every N cells
// written with collective size of 'shippedCallSizeLimit') we will call shipped which
// may clear prevBlocks list.
kvs.shipped();
shipper.shipped();
bytesWrittenProgressForShippedCall = 0;
}
if (lastCleanCell != null) {

View File

@ -0,0 +1,146 @@
/*
* Licensed to the Apache Software Foundation (ASF) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. The ASF licenses this file
* to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance
* with the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.apache.hadoop.hbase.regionserver;
import static org.hamcrest.MatcherAssert.assertThat;
import java.io.IOException;
import java.util.List;
import java.util.Optional;
import java.util.concurrent.atomic.AtomicInteger;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.hbase.Cell;
import org.apache.hadoop.hbase.HBaseClassTestRule;
import org.apache.hadoop.hbase.HBaseTestingUtil;
import org.apache.hadoop.hbase.HConstants;
import org.apache.hadoop.hbase.TableName;
import org.apache.hadoop.hbase.client.RegionLocator;
import org.apache.hadoop.hbase.client.Table;
import org.apache.hadoop.hbase.coprocessor.ObserverContext;
import org.apache.hadoop.hbase.coprocessor.RegionCoprocessor;
import org.apache.hadoop.hbase.coprocessor.RegionCoprocessorEnvironment;
import org.apache.hadoop.hbase.coprocessor.RegionObserver;
import org.apache.hadoop.hbase.regionserver.compactions.CompactionLifeCycleTracker;
import org.apache.hadoop.hbase.regionserver.compactions.CompactionRequest;
import org.apache.hadoop.hbase.testclassification.CoprocessorTests;
import org.apache.hadoop.hbase.testclassification.MediumTests;
import org.apache.hadoop.hbase.util.Bytes;
import org.hamcrest.Matchers;
import org.junit.AfterClass;
import org.junit.BeforeClass;
import org.junit.ClassRule;
import org.junit.Rule;
import org.junit.Test;
import org.junit.experimental.categories.Category;
import org.junit.rules.TestName;
@Category({ MediumTests.class, CoprocessorTests.class })
public class TestCompactionWithShippingCoprocessor {
private static final AtomicInteger SHIPPED_COUNT = new AtomicInteger();
@ClassRule
public static final HBaseClassTestRule CLASS_RULE =
HBaseClassTestRule.forClass(TestCompactionWithShippingCoprocessor.class);
protected final static HBaseTestingUtil TEST_UTIL = new HBaseTestingUtil();
private static final byte[] FAMILY = Bytes.toBytes("testFamily");
@Rule
public TestName name = new TestName();
@BeforeClass
public static void setUpBeforeClass() throws Exception {
Configuration conf = TEST_UTIL.getConfiguration();
conf.setInt(HConstants.HBASE_CLIENT_RETRIES_NUMBER, 0);// do not retry
TEST_UTIL.startMiniCluster(1);
}
@AfterClass
public static void tearDownAfterClass() throws Exception {
TEST_UTIL.shutdownMiniCluster();
}
/**
* Verifies that if a coproc returns an InternalScanner which implements Shipper, the shippped
* method is appropriately called in Compactor.
*/
@Test
public void testCoprocScannersExtendingShipperGetShipped() throws Exception {
int shippedCountBefore = SHIPPED_COUNT.get();
final TableName tableName = TableName.valueOf(name.getMethodName());
// Create a table with block size as 1024
final Table table = TEST_UTIL.createTable(tableName, new byte[][] { FAMILY }, 1, 1024,
CompactionObserver.class.getName());
TEST_UTIL.loadTable(table, FAMILY);
TEST_UTIL.flush();
try {
// get the block cache and region
RegionLocator locator = TEST_UTIL.getConnection().getRegionLocator(tableName);
String regionName = locator.getAllRegionLocations().get(0).getRegion().getEncodedName();
HRegion region = TEST_UTIL.getRSForFirstRegionInTable(tableName).getRegion(regionName);
// trigger a major compaction
TEST_UTIL.compact(true);
assertThat(SHIPPED_COUNT.get(), Matchers.greaterThan(shippedCountBefore));
} finally {
table.close();
}
}
public static class CompactionObserver implements RegionCoprocessor, RegionObserver {
@Override
public Optional<RegionObserver> getRegionObserver() {
return Optional.of(this);
}
@Override
public InternalScanner preCompact(ObserverContext<RegionCoprocessorEnvironment> c, Store store,
InternalScanner scanner, ScanType scanType, CompactionLifeCycleTracker tracker,
CompactionRequest request) throws IOException {
return new ShippedObservingScanner(scanner);
}
}
public static class ShippedObservingScanner implements InternalScanner, Shipper {
protected final InternalScanner scanner;
public ShippedObservingScanner(InternalScanner scanner) {
this.scanner = scanner;
}
@Override
public boolean next(List<Cell> result, ScannerContext scannerContext) throws IOException {
return scanner.next(result, scannerContext);
}
@Override
public void close() throws IOException {
scanner.close();
}
@Override
public void shipped() throws IOException {
if (scanner instanceof Shipper) {
SHIPPED_COUNT.incrementAndGet();
((Shipper) scanner).shipped();
}
}
}
}