HBASE-21943 The usage of RegionLocations.mergeRegionLocations is wrong for async client

Signed-off-by: Guanghao Zhang <zghao@apache.org>
This commit is contained in:
Duo Zhang 2019-02-25 16:01:38 +08:00
parent 99362e894c
commit c33ceb23d3
3 changed files with 159 additions and 55 deletions

View File

@ -23,7 +23,6 @@ import static org.apache.hadoop.hbase.TableName.META_TABLE_NAME;
import static org.apache.hadoop.hbase.client.AsyncRegionLocatorHelper.canUpdateOnError;
import static org.apache.hadoop.hbase.client.AsyncRegionLocatorHelper.createRegionLocations;
import static org.apache.hadoop.hbase.client.AsyncRegionLocatorHelper.isGood;
import static org.apache.hadoop.hbase.client.AsyncRegionLocatorHelper.mergeRegionLocations;
import static org.apache.hadoop.hbase.client.AsyncRegionLocatorHelper.removeRegionLocation;
import static org.apache.hadoop.hbase.client.ConnectionUtils.createClosestRowAfter;
import static org.apache.hadoop.hbase.client.ConnectionUtils.isEmptyStopRow;
@ -218,7 +217,7 @@ class AsyncNonMetaRegionLocator {
if (loc1.getSeqNum() != loc2.getSeqNum()) {
return false;
}
if (Objects.equal(loc1.getServerName(), loc2.getServerName())) {
if (!Objects.equal(loc1.getServerName(), loc2.getServerName())) {
return false;
}
}
@ -226,25 +225,42 @@ class AsyncNonMetaRegionLocator {
return true;
}
// return whether we add this loc to cache
private boolean addToCache(TableCache tableCache, RegionLocations locs) {
// if we successfully add the locations to cache, return the locations, otherwise return the one
// which prevents us being added. The upper layer can use this value to complete pending requests.
private RegionLocations addToCache(TableCache tableCache, RegionLocations locs) {
LOG.trace("Try adding {} to cache", locs);
byte[] startKey = locs.getDefaultRegionLocation().getRegion().getStartKey();
for (;;) {
RegionLocations oldLocs = tableCache.cache.putIfAbsent(startKey, locs);
if (oldLocs == null) {
return true;
return locs;
}
RegionLocations mergedLocs = mergeRegionLocations(locs, oldLocs);
// check whether the regions are the same, this usually happens when table is split/merged, or
// deleted and recreated again.
RegionInfo region = locs.getRegionLocation().getRegion();
RegionInfo oldRegion = oldLocs.getRegionLocation().getRegion();
if (region.getEncodedName().equals(oldRegion.getEncodedName())) {
RegionLocations mergedLocs = oldLocs.mergeLocations(locs);
if (isEqual(mergedLocs, oldLocs)) {
// the merged one is the same with the old one, give up
LOG.trace("Will not add {} to cache because the old value {} " +
" is newer than us or has the same server name." +
" Maybe it is updated before we replace it", locs, oldLocs);
return false;
return oldLocs;
}
if (tableCache.cache.replace(startKey, oldLocs, mergedLocs)) {
return true;
return mergedLocs;
}
} else {
// the region is different, here we trust the one we fetched. This maybe wrong but finally
// the upper layer can detect this and trigger removal of the wrong locations
if (LOG.isDebugEnabled()) {
LOG.debug("The newnly fetch region {} is different from the old one {} for row '{}'," +
" try replaing the old one...", region, oldRegion, Bytes.toStringBinary(startKey));
}
if (tableCache.cache.replace(startKey, oldLocs, locs)) {
return locs;
}
}
}
}
@ -258,10 +274,27 @@ class AsyncNonMetaRegionLocator {
Optional<LocateRequest> toSend = Optional.empty();
TableCache tableCache = getTableCache(tableName);
if (locs != null) {
if (!addToCache(tableCache, locs)) {
// someone is ahead of us.
RegionLocations addedLocs = addToCache(tableCache, locs);
synchronized (tableCache) {
tableCache.pendingRequests.remove(req);
tableCache.clearCompletedRequests(Optional.of(addedLocs));
// Remove a complete locate request in a synchronized block, so the table cache must have
// quota to send a candidate request.
toSend = tableCache.getCandidate();
toSend.ifPresent(r -> tableCache.send(r));
}
toSend.ifPresent(r -> locateInMeta(tableName, r));
} else {
// we meet an error
assert error != null;
synchronized (tableCache) {
tableCache.pendingRequests.remove(req);
// fail the request itself, no matter whether it is a DoNotRetryIOException, as we have
// already retried several times
CompletableFuture<?> future = tableCache.allRequests.remove(req);
if (future != null) {
future.completeExceptionally(error);
}
tableCache.clearCompletedRequests(Optional.empty());
// Remove a complete locate request in a synchronized block, so the table cache must have
// quota to send a candidate request.
@ -269,27 +302,8 @@ class AsyncNonMetaRegionLocator {
toSend.ifPresent(r -> tableCache.send(r));
}
toSend.ifPresent(r -> locateInMeta(tableName, r));
return;
}
}
synchronized (tableCache) {
tableCache.pendingRequests.remove(req);
if (error != null) {
// fail the request itself, no matter whether it is a DoNotRetryIOException, as we have
// already retried several times
CompletableFuture<?> future = tableCache.allRequests.remove(req);
if (future != null) {
future.completeExceptionally(error);
}
}
tableCache.clearCompletedRequests(Optional.ofNullable(locs));
// Remove a complete locate request in a synchronized block, so the table cache must have
// quota to send a candidate request.
toSend = tableCache.getCandidate();
toSend.ifPresent(r -> tableCache.send(r));
}
toSend.ifPresent(r -> locateInMeta(tableName, r));
}
// return whether we should stop the scan
private boolean onScanNext(TableName tableName, LocateRequest req, Result result) {
@ -443,10 +457,9 @@ class AsyncNonMetaRegionLocator {
if (info == null || info.isOffline() || info.isSplitParent()) {
continue;
}
if (addToCache(tableCache, locs)) {
RegionLocations addedLocs = addToCache(tableCache, locs);
synchronized (tableCache) {
tableCache.clearCompletedRequests(Optional.of(locs));
}
tableCache.clearCompletedRequests(Optional.of(addedLocs));
}
}
}

View File

@ -129,20 +129,6 @@ final class AsyncRegionLocatorHelper {
}
}
/**
* Create a new {@link RegionLocations} which is the merging result for the given two
* {@link RegionLocations}.
* <p/>
* All the {@link RegionLocations} in async locator related class are immutable because we want to
* access them concurrently, so here we need to create a new one, instead of calling
* {@link RegionLocations#mergeLocations(RegionLocations)} directly.
*/
static RegionLocations mergeRegionLocations(RegionLocations newLocs, RegionLocations oldLocs) {
RegionLocations locs = new RegionLocations(newLocs.getRegionLocations());
locs.mergeLocations(oldLocs);
return locs;
}
static boolean isGood(RegionLocations locs, int replicaId) {
if (locs == null) {
return false;

View File

@ -0,0 +1,105 @@
/**
* 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.client;
import static org.junit.Assert.assertFalse;
import static org.junit.Assert.assertTrue;
import java.io.IOException;
import org.apache.commons.io.IOUtils;
import org.apache.hadoop.hbase.HBaseClassTestRule;
import org.apache.hadoop.hbase.HBaseTestingUtility;
import org.apache.hadoop.hbase.TableName;
import org.apache.hadoop.hbase.testclassification.ClientTests;
import org.apache.hadoop.hbase.testclassification.MediumTests;
import org.apache.hadoop.hbase.util.Bytes;
import org.junit.AfterClass;
import org.junit.BeforeClass;
import org.junit.ClassRule;
import org.junit.Test;
import org.junit.experimental.categories.Category;
/**
* Fix an infinite loop in {@link AsyncNonMetaRegionLocator}, see the comments on HBASE-21943 for
* more details.
*/
@Category({ MediumTests.class, ClientTests.class })
public class TestAsyncTableLocateRegionForDeletedTable {
@ClassRule
public static final HBaseClassTestRule CLASS_RULE =
HBaseClassTestRule.forClass(TestAsyncTableLocateRegionForDeletedTable.class);
private static final HBaseTestingUtility TEST_UTIL = new HBaseTestingUtility();
private static TableName TABLE_NAME = TableName.valueOf("async");
private static byte[] FAMILY = Bytes.toBytes("cf");
private static byte[] QUALIFIER = Bytes.toBytes("cq");
private static byte[] VALUE = Bytes.toBytes("value");
private static AsyncConnection ASYNC_CONN;
@BeforeClass
public static void setUpBeforeClass() throws Exception {
TEST_UTIL.startMiniCluster(3);
TEST_UTIL.createTable(TABLE_NAME, FAMILY);
TEST_UTIL.waitTableAvailable(TABLE_NAME);
TEST_UTIL.getAdmin().balancerSwitch(false, true);
ASYNC_CONN = ConnectionFactory.createAsyncConnection(TEST_UTIL.getConfiguration()).get();
assertFalse(ASYNC_CONN.isClosed());
}
@AfterClass
public static void tearDownAfterClass() throws Exception {
IOUtils.closeQuietly(ASYNC_CONN);
assertTrue(ASYNC_CONN.isClosed());
TEST_UTIL.shutdownMiniCluster();
}
@Test
public void test() throws IOException, InterruptedException {
try (Table table = TEST_UTIL.getConnection().getTable(TABLE_NAME)) {
for (int i = 0; i < 100; i++) {
table.put(new Put(Bytes.toBytes(i)).addColumn(FAMILY, QUALIFIER, VALUE));
}
}
TEST_UTIL.getAdmin().split(TABLE_NAME, Bytes.toBytes(50));
TEST_UTIL.waitFor(60000,
() -> TEST_UTIL.getMiniHBaseCluster().getRegions(TABLE_NAME).size() == 2);
// make sure we can access the split regions
try (Table table = TEST_UTIL.getConnection().getTable(TABLE_NAME)) {
for (int i = 0; i < 100; i++) {
assertFalse(table.get(new Get(Bytes.toBytes(i))).isEmpty());
}
}
// let's cache the two old locations
AsyncTableRegionLocator locator = ASYNC_CONN.getRegionLocator(TABLE_NAME);
locator.getRegionLocation(Bytes.toBytes(0)).join();
locator.getRegionLocation(Bytes.toBytes(99)).join();
// recreate the table
TEST_UTIL.getAdmin().disableTable(TABLE_NAME);
TEST_UTIL.getAdmin().deleteTable(TABLE_NAME);
TEST_UTIL.createTable(TABLE_NAME, FAMILY);
TEST_UTIL.waitTableAvailable(TABLE_NAME);
// confirm that we can still get the correct location
assertFalse(ASYNC_CONN.getTable(TABLE_NAME).exists(new Get(Bytes.toBytes(99))).join());
}
}