You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@geode.apache.org by mk...@apache.org on 2022/04/04 06:51:45 UTC
[geode] branch develop updated: GEODE-9854: Orphaned .drf file causing memory leak (#7234)
This is an automated email from the ASF dual-hosted git repository.
mkevo pushed a commit to branch develop
in repository https://gitbox.apache.org/repos/asf/geode.git
The following commit(s) were added to refs/heads/develop by this push:
new 143ffb48ea GEODE-9854: Orphaned .drf file causing memory leak (#7234)
143ffb48ea is described below
commit 143ffb48ea2a28b19e4d184a68cb6484c9e0feb8
Author: Jakov Varenina <62...@users.noreply.github.com>
AuthorDate: Mon Apr 4 08:51:38 2022 +0200
GEODE-9854: Orphaned .drf file causing memory leak (#7234)
* GEODE-9854: Orphaned .drf file causing memory leak
Issue:
An OpLog files are compacted, but the .drf file is left because it contains deletes of
entries in previous .crfs. The .crf file is deleted, but the orphaned .drf is not until all
previous .crf files (.crfs with smaller id) are deleted.
The problem is that compacted Oplog object representing orphaned .drf file holds
a structure in memory (Oplog.regionMap) that contains information that is not useful
after the compaction and it takes certain amount of memory. Besides, there is a special
case when creating .krf files that, depending on the execution order,
could make the problem more severe (it could leave pendingKrfTags structure
on the regionMap and this could take up a significant amount of memory). This
pendingKrfTags HashMap is actually empty, but consumes memory because it was used
previously and the size of the HashMap was not reduced after it is cleared.
This special case usually happens when new Oplog is rolled out and previous Oplog
is immediately marked as eligible for compaction. Compaction and .krf creation start at
the similar time and compactor cancels creation of .krf file.
The pendingKrfTags structure is usually cleared when .krf file is created, but since
compaction canceled creation of .krf, the pendingKrfTags structure remain in memory
until Oplog representing orphaned .drf file is deleted.
Solution:
Clear the regionMap data structure of the Oplog when it is compacted (currently it is
deleted when the Oplog is destroyed).
Co-authored-by: Alberto Gomez <al...@est.tech>
---
...ObjectThatAreNoLongerNeededIntegrationTest.java | 449 +++++++++++++++++++++
.../org/apache/geode/internal/cache/Oplog.java | 69 +++-
2 files changed, 498 insertions(+), 20 deletions(-)
diff --git a/geode-core/src/integrationTest/java/org/apache/geode/internal/cache/DiskRegionCompactorClearsObjectThatAreNoLongerNeededIntegrationTest.java b/geode-core/src/integrationTest/java/org/apache/geode/internal/cache/DiskRegionCompactorClearsObjectThatAreNoLongerNeededIntegrationTest.java
new file mode 100644
index 0000000000..ed1e06687c
--- /dev/null
+++ b/geode-core/src/integrationTest/java/org/apache/geode/internal/cache/DiskRegionCompactorClearsObjectThatAreNoLongerNeededIntegrationTest.java
@@ -0,0 +1,449 @@
+/*
+ * 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.geode.internal.cache;
+
+import static org.apache.geode.test.awaitility.GeodeAwaitility.await;
+import static org.assertj.core.api.Assertions.assertThat;
+
+import java.io.File;
+import java.util.Arrays;
+import java.util.Properties;
+
+import org.junit.After;
+import org.junit.Before;
+import org.junit.Rule;
+import org.junit.Test;
+import org.junit.rules.TemporaryFolder;
+import org.junit.rules.TestName;
+
+import org.apache.geode.cache.Cache;
+import org.apache.geode.cache.CacheFactory;
+import org.apache.geode.cache.DataPolicy;
+import org.apache.geode.cache.Region;
+import org.apache.geode.cache.RegionFactory;
+import org.apache.geode.cache.RegionShortcut;
+
+/**
+ * Verifies that the unnecessary memory is cleared when operational log is compacted.
+ */
+public class DiskRegionCompactorClearsObjectThatAreNoLongerNeededIntegrationTest {
+
+ private final Properties config = new Properties();
+ private Cache cache;
+
+ private File[] diskDirs;
+ private int[] diskDirSizes;
+
+ private String regionName;
+ private String diskStoreName;
+
+ @Rule
+ public TemporaryFolder temporaryFolder = new TemporaryFolder();
+
+ @Rule
+ public TestName testName = new TestName();
+
+ private static final int ENTRY_RANGE_0_299 = 300;
+ private static final int ENTRY_RANGE_300_599 = 600;
+
+ @Before
+ public void setUp() throws Exception {
+ String uniqueName = getClass().getSimpleName() + "_" + testName.getMethodName();
+ regionName = uniqueName + "_region";
+ diskStoreName = uniqueName + "_diskStore";
+
+ cache = new CacheFactory(config).create();
+
+ diskDirs = new File[1];
+ diskDirs[0] = createDirectory(temporaryFolder.getRoot(), testName.getMethodName());
+ diskDirSizes = new int[1];
+ Arrays.fill(diskDirSizes, Integer.MAX_VALUE);
+
+ DiskStoreImpl.SET_IGNORE_PREALLOCATE = true;
+ TombstoneService.EXPIRED_TOMBSTONE_LIMIT = 1;
+ TombstoneService.REPLICATE_TOMBSTONE_TIMEOUT = 1;
+ }
+
+ @After
+ public void tearDown() throws Exception {
+ try {
+ cache.close();
+ } finally {
+ DiskStoreImpl.SET_IGNORE_PREALLOCATE = false;
+ }
+ }
+
+ /**
+ * Verifies that the unnecessary memory is cleared when operational log (.crf and .drf) is
+ * compacted.
+ * This test case covers the following scenario:
+ *
+ * 1. Create several Oplog files (.crf, .drf and .krf) by executing put operations
+ * 2. Execute destroy operation for every fifth entry, and each time add new entry. This will
+ * result with few additional Oplog files. Compaction threshold will not be reached.
+ * 3. Destroy all operations created in step 2. This will trigger compaction of files that
+ * were created in step 2. Compaction will delete only .crf and .krf files, but will not
+ * delete .drf files because they contain destroy operations for events located in
+ * .crf files crated in step 1.
+ * 4. Check that unnecessary objects are cleared for theOplog that represents orphaned .drf
+ * file (no accompanying .crf and .krf file)
+ **/
+ @Test
+ public void testCompactorRegionMapDeletedForOnlyDrfOplogAfterCompactionIsDone()
+ throws InterruptedException {
+
+ createDiskStore(30, 10000);
+ Region<Object, Object> region = createRegion();
+ DiskStoreImpl diskStore = ((InternalRegion) region).getDiskStore();
+
+ // Create several oplog files (.crf and .drf) by executing put operations in defined range
+ executePutsInRange0_299(region);
+ await().untilAsserted(() -> assertThat(getCurrentNumberOfOplogs(diskStore)).isEqualTo(5));
+
+ // Destroy every fifth entry from previous range and each time put new entry in new range.
+ // This will create additional oplog files (.crf and .drf), but compaction will not be triggered
+ // as threshold will not be reached. Oplog files (.drf) created in this step will contain
+ // destroys for events that are located in .crf files from previous range.
+ destroyEveryFifthElementInRange0_299AndEachTimePutInRange300_599(region);
+ await().untilAsserted(() -> assertThat(getCurrentNumberOfOplogs(diskStore)).isEqualTo(7));
+
+ // Destroy all events created in previous step in order to trigger automatic compaction.
+ // This will trigger compaction for the files that were created in previous step.
+ // Compaction will delete .crf and .krf file, but will leave .drf file because it contains
+ // destroy operation for the events that are located in some older .crf files.
+ destroyEveryFifthElementInRange300_599(region);
+
+ // wait for all Oplog's to be compacted
+ await().untilAsserted(() -> assertThat(isOplogToBeCompactedAvailable(diskStore)).isFalse());
+
+ await().untilAsserted(
+ () -> assertThat(areAllUnnecessaryObjectClearedForOnlyDrfOplog(diskStore)).isTrue());
+ }
+
+ /**
+ * Verifies that the unnecessary memory is cleared when operational log (.crf and .drf) is
+ * compacted. This test case covers the following scenario:
+ *
+ * 1. Create several Oplog files (.crf, .drf and .krf) by executing put operations
+ * 2. Execute destroy operation for every fifth entry from previous step, and each time add and
+ * destroy new entry. This will result with Oplogs that have 0 live entries at the moment when
+ * they are rolled out. These Oplogs will be closed early since compaction doesn't have to be
+ * done for them. When Oplog is closed early the .crf files will be deleted, but .drf files will
+ * not because they contain deletes.
+ * 3. Check that unnecessary objects are cleared for theOplog that represents orphaned .drf
+ * file (no accompanying .crf and .krf file)
+ **/
+ @Test
+ public void testCompactorRegionMapDeletedForOnlyDrfOplogAfterCompactionNotDoneDueToLiveCountZero()
+ throws InterruptedException {
+
+ createDiskStore(5, 10000);
+ Region<Object, Object> region = createRegion();
+ DiskStoreImpl diskStore = ((InternalRegion) region).getDiskStore();
+
+ // Create several oplog files (.crf and .drf) by executing put operations in defined range
+ executePutsInRange0_299(region);
+ await().untilAsserted(() -> assertThat(getCurrentNumberOfOplogs(diskStore)).isEqualTo(5));
+
+ // Do multiple put and destroy operations to hit the scenario in which there are 0
+ // live entries are in Oplog just before compaction is triggered.
+ destroyEveryFifthElementInRange0_299PutAndDestroyInRange300_599(region);
+
+ await().untilAsserted(() -> assertThat(isOplogToBeCompactedAvailable(diskStore)).isFalse());
+
+ await().untilAsserted(
+ () -> assertThat(areAllUnnecessaryObjectClearedForOnlyDrfOplog(diskStore)).isTrue());
+ }
+
+ /**
+ * Verifies that the unnecessary memory is cleared when operational log (.crf and .drf) is
+ * compacted.This is special scenario were creation of .krf file is cancelled by ongoing
+ * compaction. This usually happens when new oplog is rolled out and previous oplog is
+ * immediately marked as eligible for compaction. Compaction and .krf creation start at the
+ * similar time and compactor cancels creation of .krf if it is executed first.
+ *
+ * This test case covers the following scenario:
+ *
+ * 1. Create several Oplog files (.crf, .drf and .krf) by executing put operations.
+ * 2. Execute destroy operation for every fifth entry, and each time add new entry. When
+ * it is time for oplog to roll out, the previous oplog will be immediately marked as ready
+ * for compaction because compaction threshold is set to high value in this case. This way
+ * we force that compaction cancel creation of .krf file. Compaction will
+ * delete only .crf file (.krf was not created at all), but will not delete .drf files because
+ * they contain destroy operations for events located in .crf files created in step 1.
+ * 3. Check that unnecessary objects are cleared for theOplog that represents orphaned .drf
+ * file (no accompanying .crf and .krf file)
+ **/
+ @Test
+ public void testCompactorRegionMapDeletedAfterCompactionForOnlyDrfOplogAndKrfCreationCanceledByCompactionIsDone() {
+
+ createDiskStore(70, 10000);
+ Region<Object, Object> region = createRegion();
+ DiskStoreImpl diskStore = ((InternalRegion) region).getDiskStore();
+
+ // Create several oplog files (.crf and .drf) by executing put operations in defined range
+ executePutsInRange0_299(region);
+ await().untilAsserted(() -> assertThat(getCurrentNumberOfOplogs(diskStore)).isEqualTo(5));
+
+ destroyEveryFifthElementInRange0_299AndEachTimePutInRange300_599(region);
+ await().untilAsserted(() -> assertThat(isOplogToBeCompactedAvailable(diskStore)).isFalse());
+
+ await().untilAsserted(
+ () -> assertThat(areAllUnnecessaryObjectClearedForOnlyDrfOplog(diskStore)).isTrue());
+ }
+
+ /**
+ * Verifies that the region is recovered from Oplog's (including .drf only oplog's) when region is
+ * closed and then recreated again in order to trigger recovery.
+ * This test case covers the following scenario:
+ *
+ * 1. Create several Oplog files (.crf, .drf and .krf) by executing put operations
+ * 2. Execute destroy operation for every fifth entry, and each time add new entry. This will
+ * result with few additional Oplog files. Compaction threshold will not be reached.
+ * 3. Destroy all operations created in step 2. This will trigger compaction of files that
+ * were created in step 2. Compaction will delete only .crf and .krf files, but will not
+ * delete .drf files because they contain destroy operations for events located in
+ * .crf files created in step 1. Check that unnecessary objects are cleared for the
+ * Oplog that represents orphaned .drf file (no accompanying .crf and .krf file)
+ * 4. At this step there will be Oplog objects from step 1. that will contain all files (.crf,
+ * .drf and .krf), and Oplog objects from step 2. that will contain only .drf files.
+ * In order to test recovery, close the region and then recreate it again.
+ * 5. Check that region is recovered correctly. Check that only events that have never been
+ * destroyed are recovered from Oplog files.
+ * 6. Check that unnecessary objects are cleared for theOplog that represents orphaned .drf
+ * file (no accompanying .crf and .krf file)
+ **/
+ @Test
+ public void testCompactorRegionMapDeletedForOnlyDrfOplogAfterCompactionAndRecoveryAfterRegionClose()
+ throws InterruptedException {
+
+ createDiskStore(30, 10000);
+ Region<Object, Object> region = createRegion();
+ DiskStoreImpl diskStore = ((InternalRegion) region).getDiskStore();
+
+ // Create several oplog files (.crf and .drf) by executing put operations in defined range
+ executePutsInRange0_299(region);
+ await().untilAsserted(() -> assertThat(getCurrentNumberOfOplogs(diskStore)).isEqualTo(5));
+
+ // Destroy every fifth entry from previous range and each time put new entry in new range.
+ // This will create additional oplog files (.crf and .drf), but compaction will not be triggered
+ // as threshold will not be reached. Oplog files (.drf) created in this step will contain
+ // destroys for events that are located in .crf files from previous range.
+ destroyEveryFifthElementInRange0_299AndEachTimePutInRange300_599(region);
+ await().untilAsserted(() -> assertThat(getCurrentNumberOfOplogs(diskStore)).isEqualTo(7));
+
+ // Destroy all events created in previous step in order to trigger automatic compaction.
+ // This will trigger compaction for the files that were created in previous step.
+ // Compaction will delete .crf and .krf file, but will leave .drf file because it contains
+ // destroy operation for the events that are located in some older .crf files.
+ destroyEveryFifthElementInRange300_599(region);
+
+ // wait for all Oplog's to be compacted
+ await().untilAsserted(() -> assertThat(isOplogToBeCompactedAvailable(diskStore)).isFalse());
+
+ // close the region and then recreate it to trigger recovery from oplog files
+ region.close();
+ region = createRegion();
+
+ // every fifth element is destroyed from range ENTRY_RANGE_1, so it is expected to have
+ // ENTRY_RANGE_1 - (ENTRY_RANGE_1/5) of elements, also just in case check that every
+ // fifth get operation return null
+ checkThatAllRegionDataIsRecoveredFromOplogFiles(region);
+
+ // check that unnecessary data is cleared from Oplog's
+ await().untilAsserted(
+ () -> assertThat(areAllUnnecessaryObjectClearedForOnlyDrfOplog(diskStore)).isTrue());
+ }
+
+ /**
+ * Verifies that the region is recovered from Oplog's (including .drf only oplog's) when cache is
+ * closed and then recreated again in order to trigger recovery.
+ * This test case covers the following scenario:
+ *
+ * 1. Create several Oplog files (.crf, .drf and .krf) by executing put operations
+ * 2. Execute destroy operation for every fifth entry, and each time add new entry. This will
+ * result with few additional Oplog files. Compaction threshold will not be reached.
+ * 3. Destroy all operations created in step 2. This will trigger compaction of files that
+ * were created in step 2. Compaction will delete only .crf and .krf files, but will not
+ * delete .drf files because they contain destroy operations for events located in
+ * .crf files created in step 1.
+ * 4. At this step there will be Oplog objects from step 1. that will contain all files (.crf,
+ * .drf and .krf), and Oplog objects from step 2. that will contain only .drf files.
+ * In order to test recovery, close the cache and then recreate it again.
+ * 5. Check that region is recovered correctly. Check that only events that have never been
+ * destroyed are recovered from Oplog files.
+ * 6. Check that unnecessary objects are cleared for theOplog that represents orphaned .drf
+ * file (no accompanying .crf and .krf file)
+ **/
+ @Test
+ public void testCompactorRegionMapDeletedForOnlyDrfOplogAfterCompactionAndRecoveryAfterCacheClosed()
+ throws InterruptedException {
+
+ createDiskStore(30, 10000);
+ Region<Object, Object> region = createRegion();
+ DiskStoreImpl diskStore = ((InternalRegion) region).getDiskStore();
+
+ // Create several oplog files (.crf and .drf) by executing put operations in defined range
+ executePutsInRange0_299(region);
+ await().untilAsserted(() -> assertThat(getCurrentNumberOfOplogs(diskStore)).isEqualTo(5));
+
+ // Destroy every fifth entry from previous range and each time put new entry in new range.
+ // This will create additional oplog files (.crf and .drf), but compaction will not be triggered
+ // as threshold will not be reached. Oplog files (.drf) created in this step will contain
+ // destroys for events that are located in .crf files from previous range.
+ destroyEveryFifthElementInRange0_299AndEachTimePutInRange300_599(region);
+ await().untilAsserted(() -> assertThat(getCurrentNumberOfOplogs(diskStore)).isEqualTo(7));
+
+ // Destroy all events created in previous step in order to trigger automatic compaction.
+ // This will trigger compaction for the files that were created in previous step.
+ // Compaction will delete .crf and .krf file, but will leave .drf file because it contains
+ // destroy operation for the events that are located in some older .crf files.
+ destroyEveryFifthElementInRange300_599(region);
+
+ // wait for all Oplog's to be compacted
+ await().untilAsserted(() -> assertThat(isOplogToBeCompactedAvailable(diskStore)).isFalse());
+
+ // close the cache and then recreate it again to trigger recovery from oplog files
+ cache.close();
+ cache = new CacheFactory(config).create();
+ createDiskStore(30, 10000);
+ region = createRegion();
+
+ // every fifth element is destroyed from range ENTRY_RANGE_1, so it is expected to have
+ // ENTRY_RANGE_1 - (ENTRY_RANGE_1/5) of elements, also just in case check that every
+ // fifth get operation return null
+ checkThatAllRegionDataIsRecoveredFromOplogFiles(region);
+
+ await().untilAsserted(
+ () -> assertThat(areAllUnnecessaryObjectClearedForOnlyDrfOplog(diskStore)).isTrue());
+ }
+
+ void checkThatAllRegionDataIsRecoveredFromOplogFiles(Region<Object, Object> region) {
+ int objectCount = 0;
+ for (int i = 0; i < ENTRY_RANGE_0_299; i++) {
+ Object object = region.get(i);
+ if (i % 5 == 0) {
+ assertThat(object).isNull();
+ }
+ if (object != null) {
+ objectCount++;
+ }
+ }
+ assertThat(objectCount).isEqualTo(ENTRY_RANGE_0_299 - (ENTRY_RANGE_0_299 / 5));
+ }
+
+ void executePutsInRange0_299(Region<Object, Object> region) {
+ for (int i = 0; i < ENTRY_RANGE_0_299; i++) {
+ region.put(i, new byte[100]);
+ }
+ }
+
+ void destroyEveryFifthElementInRange0_299PutAndDestroyInRange300_599(
+ Region<Object, Object> region)
+ throws InterruptedException {
+ TombstoneService tombstoneService = ((InternalCache) cache).getTombstoneService();
+ for (int i = ENTRY_RANGE_0_299; i < ENTRY_RANGE_300_599; i++) {
+ region.put(i, new byte[300]);
+ region.destroy(i);
+ assertThat(tombstoneService.forceBatchExpirationForTests(1)).isTrue();
+ if (i % 5 == 0) {
+ region.destroy(i - ENTRY_RANGE_0_299);
+ assertThat(tombstoneService.forceBatchExpirationForTests(1)).isTrue();
+ }
+ }
+ }
+
+ void destroyEveryFifthElementInRange300_599(Region<Object, Object> region)
+ throws InterruptedException {
+ TombstoneService tombstoneService = ((InternalCache) cache).getTombstoneService();
+ int key = ENTRY_RANGE_0_299;
+ while (key < ENTRY_RANGE_300_599) {
+ region.destroy(key);
+ assertThat(tombstoneService.forceBatchExpirationForTests(1)).isTrue();
+ key = key + 5;
+ }
+ }
+
+ void destroyEveryFifthElementInRange0_299AndEachTimePutInRange300_599(
+ Region<Object, Object> region) {
+ TombstoneService tombstoneService = ((InternalCache) cache).getTombstoneService();
+ int key = 0;
+ while (key < ENTRY_RANGE_0_299) {
+ region.destroy(key);
+ // It is necessary to force tombstone expiration, otherwise event won't be stored in .drf file
+ // and total live count won't be decreased
+ await().untilAsserted(
+ () -> assertThat(tombstoneService.forceBatchExpirationForTests(1)).isTrue());
+ region.put(key + ENTRY_RANGE_0_299, new byte[300]);
+ key = key + 5;
+ }
+ }
+
+ boolean areAllUnnecessaryObjectClearedForOnlyDrfOplog(DiskStoreImpl diskStore) {
+ boolean isClear = true;
+ for (Oplog oplog : diskStore.getAllOplogsForBackup()) {
+ if (oplog.getHasDeletes() && oplog.isDeleted() && oplog.hasNoLiveValues()) {
+ if (oplog.getRegionMapSize() != 0)
+ isClear = false;
+ }
+ }
+ return isClear;
+ }
+
+ void createDiskStore(int compactionThreshold, int maxOplogSizeInBytes) {
+ DiskStoreFactoryImpl diskStoreFactory = (DiskStoreFactoryImpl) cache.createDiskStoreFactory();
+ diskStoreFactory.setAutoCompact(true);
+ diskStoreFactory.setCompactionThreshold(compactionThreshold);
+ diskStoreFactory.setDiskDirsAndSizes(diskDirs, diskDirSizes);
+
+ createDiskStoreWithSizeInBytes(diskStoreName, diskStoreFactory, maxOplogSizeInBytes);
+ }
+
+ Region<Object, Object> createRegion() {
+ RegionFactory<Object, Object> regionFactory =
+ cache.createRegionFactory(RegionShortcut.PARTITION_PERSISTENT);
+ regionFactory.setDataPolicy(DataPolicy.PERSISTENT_PARTITION);
+ regionFactory.setDiskStoreName(diskStoreName);
+ regionFactory.setDiskSynchronous(true);
+ return regionFactory.create(regionName);
+ }
+
+ boolean isOplogToBeCompactedAvailable(DiskStoreImpl ds) {
+ if (ds.getOplogToBeCompacted() == null) {
+ return false;
+ }
+ return ds.getOplogToBeCompacted().length > 0;
+ }
+
+ int getCurrentNumberOfOplogs(DiskStoreImpl ds) {
+ return ds.getAllOplogsForBackup().length;
+ }
+
+ private File createDirectory(File parentDirectory, String name) {
+ File file = new File(parentDirectory, name);
+ assertThat(file.mkdir()).isTrue();
+ return file;
+ }
+
+ private void createDiskStoreWithSizeInBytes(String diskStoreName,
+ DiskStoreFactoryImpl diskStoreFactory,
+ long maxOplogSizeInBytes) {
+ diskStoreFactory.setMaxOplogSizeInBytes(maxOplogSizeInBytes);
+ diskStoreFactory.setDiskDirSizesUnit(DiskDirSizesUnit.BYTES);
+ diskStoreFactory.create(diskStoreName);
+ }
+}
diff --git a/geode-core/src/main/java/org/apache/geode/internal/cache/Oplog.java b/geode-core/src/main/java/org/apache/geode/internal/cache/Oplog.java
index 554b43c168..7d8737040b 100755
--- a/geode-core/src/main/java/org/apache/geode/internal/cache/Oplog.java
+++ b/geode-core/src/main/java/org/apache/geode/internal/cache/Oplog.java
@@ -44,6 +44,7 @@ import java.util.concurrent.ConcurrentMap;
import java.util.concurrent.atomic.AtomicBoolean;
import java.util.concurrent.atomic.AtomicInteger;
import java.util.concurrent.atomic.AtomicLong;
+import java.util.concurrent.atomic.AtomicReference;
import java.util.concurrent.locks.Lock;
import java.util.concurrent.locks.ReentrantLock;
@@ -191,9 +192,6 @@ public class Oplog implements CompactableOplog, Flushable {
*/
private final AtomicLong totalLiveCount = new AtomicLong(0);
- private final ConcurrentMap<Long, DiskRegionInfo> regionMap =
- new ConcurrentHashMap<>();
-
/**
* Set to true once compact is called on this oplog.
*
@@ -529,6 +527,11 @@ public class Oplog implements CompactableOplog, Flushable {
*/
private boolean doneAppending = false;
+ /**
+ * Used to track all information's about live entries that region has in this oplog.
+ */
+ private final RegionMap regionMap = new RegionMap();
+
/**
* Creates new {@code Oplog} for the given region.
*
@@ -811,7 +814,7 @@ public class Oplog implements CompactableOplog, Flushable {
* Return true if this oplog has a drf but does not have a crf
*/
boolean isDrfOnly() {
- return drf.f != null && crf.f == null;
+ return drf.f != null && (crf.f == null || !crf.f.exists());
}
/**
@@ -1302,17 +1305,19 @@ public class Oplog implements CompactableOplog, Flushable {
// while a krf is being created can not close a region
lockCompactor();
try {
- DiskRegionInfo dri = getDRI(dr);
- if (dri != null) {
- long clearCount = dri.clear(null);
- if (clearCount != 0) {
- totalLiveCount.addAndGet(-clearCount);
- // no need to call handleNoLiveValues because we now have an
- // unrecovered region.
+ if (!isDrfOnly()) {
+ DiskRegionInfo dri = getDRI(dr);
+ if (dri != null) {
+ long clearCount = dri.clear(null);
+ if (clearCount != 0) {
+ totalLiveCount.addAndGet(-clearCount);
+ // no need to call handleNoLiveValues because we now have an
+ // unrecovered region.
+ }
+ regionMap.get().remove(dr.getId(), dri);
}
- regionMap.remove(dr.getId(), dri);
+ addUnrecoveredRegion(dr.getId());
}
- addUnrecoveredRegion(dr.getId());
} finally {
unlockCompactor();
}
@@ -1341,7 +1346,7 @@ public class Oplog implements CompactableOplog, Flushable {
handleNoLiveValues();
}
}
- regionMap.remove(dr.getId(), dri);
+ regionMap.get().remove(dr.getId(), dri);
}
}
@@ -4010,7 +4015,7 @@ public class Oplog implements CompactableOplog, Flushable {
return;
}
- Collection<DiskRegionInfo> regions = regionMap.values();
+ Collection<DiskRegionInfo> regions = regionMap.get().values();
List<KRFEntry> sortedLiveEntries = getSortedLiveEntries(regions);
if (sortedLiveEntries == null) {
// no need to create a krf if there are no live entries.
@@ -4670,7 +4675,7 @@ public class Oplog implements CompactableOplog, Flushable {
}
private DiskRegionInfo getDRI(long drId) {
- return regionMap.get(drId);
+ return regionMap.get().get(drId);
}
private DiskRegionInfo getDRI(DiskRegionView dr) {
@@ -4683,7 +4688,7 @@ public class Oplog implements CompactableOplog, Flushable {
dri = (isCompactionPossible() || couldHaveKrf())
? new DiskRegionInfoWithList(dr, couldHaveKrf(), krfCreated.get())
: new DiskRegionInfoNoList(dr);
- DiskRegionInfo oldDri = regionMap.putIfAbsent(dr.getId(), dri);
+ DiskRegionInfo oldDri = regionMap.get().putIfAbsent(dr.getId(), dri);
if (oldDri != null) {
dri = oldDri;
}
@@ -4708,7 +4713,7 @@ public class Oplog implements CompactableOplog, Flushable {
dri = (isCompactionPossible() || couldHaveKrf())
? new DiskRegionInfoWithList(null, couldHaveKrf(), krfCreated.get())
: new DiskRegionInfoNoList(null);
- DiskRegionInfo oldDri = regionMap.putIfAbsent(drId, dri);
+ DiskRegionInfo oldDri = regionMap.get().putIfAbsent(drId, dri);
if (oldDri != null) {
dri = oldDri;
}
@@ -5714,6 +5719,7 @@ public class Oplog implements CompactableOplog, Flushable {
cancelKrf();
close();
deleteFiles(getHasDeletes());
+ regionMap.close();
} finally {
unlockCompactor();
}
@@ -5861,7 +5867,7 @@ public class Oplog implements CompactableOplog, Flushable {
* getParent().getOwner().isDestroyed ||
*/!compactor.keepCompactorRunning();
int totalCount = 0;
- for (DiskRegionInfo dri : regionMap.values()) {
+ for (DiskRegionInfo dri : regionMap.get().values()) {
final DiskRegionView dr = dri.getDiskRegion();
if (dr == null) {
continue;
@@ -5976,7 +5982,7 @@ public class Oplog implements CompactableOplog, Flushable {
List<KRFEntry> sortedLiveEntries;
- HashMap<Long, DiskRegionInfo> targetRegions = new HashMap<>(regionMap);
+ HashMap<Long, DiskRegionInfo> targetRegions = new HashMap<>(regionMap.get());
synchronized (diskRecoveryStores) {
diskRecoveryStores.values()
.removeIf(store -> isLruValueRecoveryDisabled(store) || store.lruLimitExceeded());
@@ -6271,6 +6277,10 @@ public class Oplog implements CompactableOplog, Flushable {
return version;
}
+ public int getRegionMapSize() {
+ return regionMap.get().size();
+ }
+
public enum OPLOG_TYPE {
CRF(new byte[] {0x47, 0x46, 0x43, 0x52, 0x46, 0x31}), // GFCRF1
DRF(new byte[] {0x47, 0x46, 0x44, 0x52, 0x46, 0x31}), // GFDRF1
@@ -7642,4 +7652,23 @@ public class Oplog implements CompactableOplog, Flushable {
}
+ /**
+ * Used to track all information's about live entries that region has in this oplog.
+ * That information is only needed until oplog is compacted. This is because compaction will
+ * clear all live entries from this oplog.
+ */
+ private static class RegionMap {
+
+ final AtomicReference<ConcurrentMap<Long, DiskRegionInfo>> regionMap =
+ new AtomicReference<>(new ConcurrentHashMap<>());
+
+ public void close() {
+ regionMap.set(null);
+ }
+
+ public ConcurrentMap<Long, DiskRegionInfo> get() {
+ ConcurrentMap<Long, DiskRegionInfo> regionConcurrentMap = regionMap.get();
+ return regionConcurrentMap != null ? regionConcurrentMap : new ConcurrentHashMap<>();
+ }
+ }
}