[LUCENE-7919] excessive use of notifyAll

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 6.6
• Fix Version/s: 7.0
• Component/s: core/index
• Lucene Fields: New
• Assignee: Unassigned
• Reporter: Guoqiang Jiang
• Created: 05/Aug/17 09:18
• Updated: 28/Aug/22 15:17
• Resolved: 07/Aug/17 22:56

Description

I am using Elasticsearch and have a write heavy scene. When tuning with jstack, I found a significant proportion of thread stacks similar to the followings:

"elasticsearch[test][bulk][T#23]" #126 daemon prio=5 os_prio=0 tid=0x00007f68f8040000 nid=0x6b1 runnable [0x00007f6918ce9000]
   java.lang.Thread.State: RUNNABLE
        at java.lang.Object.notifyAll(Native Method)
        at org.apache.lucene.index.DocumentsWriterPerThreadPool.release(DocumentsWriterPerThreadPool.java:213)
        - locked <0x00000000ea02b6d0> (a org.apache.lucene.index.DocumentsWriterPerThreadPool)
        at org.apache.lucene.index.DocumentsWriter.updateDocument(DocumentsWriter.java:496)
        at org.apache.lucene.index.IndexWriter.updateDocument(IndexWriter.java:1571)
        at org.apache.lucene.index.IndexWriter.addDocument(IndexWriter.java:1316)
        at org.elasticsearch.index.engine.InternalEngine.index(InternalEngine.java:663)
        at org.elasticsearch.index.engine.InternalEngine.indexIntoLucene(InternalEngine.java:607)
        at org.elasticsearch.index.engine.InternalEngine.index(InternalEngine.java:505)
        at org.elasticsearch.index.shard.IndexShard.index(IndexShard.java:556)
        at org.elasticsearch.index.shard.IndexShard.index(IndexShard.java:545)
        at org.elasticsearch.action.bulk.TransportShardBulkAction.executeIndexRequestOnPrimary(TransportShardBulkAction.java:484)
        at org.elasticsearch.action.bulk.TransportShardBulkAction.executeBulkItemRequest(TransportShardBulkAction.java:143)
        at org.elasticsearch.action.bulk.TransportShardBulkAction.shardOperationOnPrimary(TransportShardBulkAction.java:113)
        at org.elasticsearch.action.bulk.TransportShardBulkAction.shardOperationOnPrimary(TransportShardBulkAction.java:69)
        at org.elasticsearch.action.support.replication.TransportReplicationAction$PrimaryShardReference.perform(TransportReplicationAction.java:939)
        at org.elasticsearch.action.support.replication.TransportReplicationAction$PrimaryShardReference.perform(TransportReplicationAction.java:908)
        at org.elasticsearch.action.support.replication.ReplicationOperation.execute(ReplicationOperation.java:113)
        at org.elasticsearch.action.support.replication.TransportReplicationAction$AsyncPrimaryAction.onResponse(TransportReplicationAction.java:322)
        at org.elasticsearch.action.support.replication.TransportReplicationAction$AsyncPrimaryAction.onResponse(TransportReplicationAction.java:264)
        at org.elasticsearch.action.support.replication.TransportReplicationAction$1.onResponse(TransportReplicationAction.java:888)
        at org.elasticsearch.action.support.replication.TransportReplicationAction$1.onResponse(TransportReplicationAction.java:885)
        at org.elasticsearch.index.shard.IndexShardOperationsLock.acquire(IndexShardOperationsLock.java:147)
        at org.elasticsearch.index.shard.IndexShard.acquirePrimaryOperationLock(IndexShard.java:1657)
        at org.elasticsearch.action.support.replication.TransportReplicationAction.acquirePrimaryShardReference(TransportReplicationAction.java:897)
        at org.elasticsearch.action.support.replication.TransportReplicationAction.access$400(TransportReplicationAction.java:93)
        at org.elasticsearch.action.support.replication.TransportReplicationAction$AsyncPrimaryAction.doRun(TransportReplicationAction.java:281)
        at org.elasticsearch.common.util.concurrent.AbstractRunnable.run(AbstractRunnable.java:37)
        at org.elasticsearch.action.support.replication.TransportReplicationAction$PrimaryOperationTransportHandler.messageReceived(TransportReplicationAction.java:260)
        at org.elasticsearch.action.support.replication.TransportReplicationAction$PrimaryOperationTransportHandler.messageReceived(TransportReplicationAction.java:252)
        at org.elasticsearch.transport.RequestHandlerRegistry.processMessageReceived(RequestHandlerRegistry.java:69)
        at org.elasticsearch.transport.TransportService$7.doRun(TransportService.java:644)
        at org.elasticsearch.common.util.concurrent.ThreadContext$ContextPreservingAbstractRunnable.doRun(ThreadContext.java:638)
        at org.elasticsearch.common.util.concurrent.AbstractRunnable.run(AbstractRunnable.java:37)
        at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)

After reading the code related with DocumentsWriterPerThreadPool, I think the notifyAll is useless. This is a relatively expensive operation, and should be avoided if possible.