Avoid background sync on relocated primary (#40800)

There were some test failures caused by the background retention lease sync running on a relocated primary. This commit fixes the situation that triggered the assertion and reactivates the failing test. Closes #40731
elastic · Apr 3, 2019 · b1c6f8e · b1c6f8e
1 parent b100f04
commit b1c6f8e
Show file tree

Hide file tree

Showing 3 changed files with 32 additions and 3 deletions.
diff --git a/qa/rolling-upgrade/src/test/java/org/elasticsearch/upgrades/RecoveryIT.java b/qa/rolling-upgrade/src/test/java/org/elasticsearch/upgrades/RecoveryIT.java
@@ -121,7 +121,6 @@ protected void doRun() throws Exception {
         return future;
     }
 
-    @AwaitsFix(bugUrl = "https://github.com/elastic/elasticsearch/issues/40731")
     public void testRecoveryWithConcurrentIndexing() throws Exception {
         final String index = "recovery_with_concurrent_indexing";
         Response response = client().performRequest(new Request("GET", "_nodes"));

diff --git a/server/src/main/java/org/elasticsearch/index/IndexService.java b/server/src/main/java/org/elasticsearch/index/IndexService.java
@@ -830,7 +830,11 @@ private void sync(final Consumer<IndexShard> sync, final String source) {
                     case STARTED:
                         try {
                             shard.runUnderPrimaryPermit(
-                                    () -> sync.accept(shard),
+                                    () -> {
+                                        if (shard.isRelocatedPrimary() == false) {
+                                            sync.accept(shard);
+                                        }
+                                    },
                                     e -> {
                                         if (e instanceof AlreadyClosedException == false
                                                 && e instanceof IndexShardClosedException == false) {

diff --git a/server/src/test/java/org/elasticsearch/recovery/RecoveryWhileUnderLoadIT.java b/server/src/test/java/org/elasticsearch/recovery/RecoveryWhileUnderLoadIT.java
@@ -30,20 +30,28 @@
 import org.elasticsearch.cluster.routing.ShardRouting;
 import org.elasticsearch.cluster.service.ClusterService;
 import org.elasticsearch.common.Priority;
+import org.elasticsearch.common.settings.Setting;
 import org.elasticsearch.common.settings.Settings;
 import org.elasticsearch.common.unit.TimeValue;
+import org.elasticsearch.index.IndexService;
 import org.elasticsearch.index.IndexSettings;
 import org.elasticsearch.index.shard.DocsStats;
 import org.elasticsearch.index.shard.ShardId;
 import org.elasticsearch.index.translog.Translog;
+import org.elasticsearch.plugins.Plugin;
 import org.elasticsearch.search.sort.SortOrder;
 import org.elasticsearch.test.BackgroundIndexer;
 import org.elasticsearch.test.ESIntegTestCase;
 import org.elasticsearch.test.junit.annotations.TestLogging;
 
 import java.util.Arrays;
+import java.util.Collection;
+import java.util.Collections;
+import java.util.List;
 import java.util.Set;
 import java.util.concurrent.TimeUnit;
+import java.util.stream.Collectors;
+import java.util.stream.Stream;
 
 import static org.elasticsearch.cluster.metadata.IndexMetaData.SETTING_NUMBER_OF_REPLICAS;
 import static org.elasticsearch.cluster.metadata.IndexMetaData.SETTING_NUMBER_OF_SHARDS;
@@ -58,6 +66,23 @@
 public class RecoveryWhileUnderLoadIT extends ESIntegTestCase {
     private final Logger logger = LogManager.getLogger(RecoveryWhileUnderLoadIT.class);
 
+    public static final class RetentionLeaseSyncIntervalSettingPlugin extends Plugin {
+
+        @Override
+        public List<Setting<?>> getSettings() {
+            return Collections.singletonList(IndexService.RETENTION_LEASE_SYNC_INTERVAL_SETTING);
+        }
+
+    }
+
+    @Override
+    protected Collection<Class<? extends Plugin>> nodePlugins() {
+        return Stream.concat(
+            super.nodePlugins().stream(),
+            Stream.of(RetentionLeaseSyncIntervalSettingPlugin.class))
+            .collect(Collectors.toList());
+    }
+
     public void testRecoverWhileUnderLoadAllocateReplicasTest() throws Exception {
         logger.info("--> creating test index ...");
         int numberOfShards = numberOfShards();
@@ -260,7 +285,8 @@ public void testRecoverWhileRelocating() throws Exception {
         assertAcked(prepareCreate("test", 3, Settings.builder()
                 .put(SETTING_NUMBER_OF_SHARDS, numShards)
                 .put(SETTING_NUMBER_OF_REPLICAS, numReplicas)
-                .put(IndexSettings.INDEX_TRANSLOG_DURABILITY_SETTING.getKey(), Translog.Durability.ASYNC)));
+                .put(IndexSettings.INDEX_TRANSLOG_DURABILITY_SETTING.getKey(), Translog.Durability.ASYNC)
+                .put(IndexService.RETENTION_LEASE_SYNC_INTERVAL_SETTING.getKey(), randomFrom("100ms", "1s", "5s", "30s", "60s"))));
 
         final int numDocs = scaledRandomIntBetween(200, 9999);