Mailing-List: contact dev-help@zookeeper.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@zookeeper.apache.org
Date: Mon, 13 Mar 2017 15:37:41 +0000 (UTC)
From: "Josh Elser (JIRA)" <jira@apache.org>
To: dev@zookeeper.apache.org
Message-ID: <JIRA.13048967.1488910744000.2218.1489419461875@Atlassian.JIRA>
In-Reply-To: <JIRA.13048967.1488910744000@Atlassian.JIRA>
References: <JIRA.13048967.1488910744000@Atlassian.JIRA> <JIRA.13048967.1488910744197@jira-lw-us.apache.org>
Subject: [jira] [Assigned] (ZOOKEEPER-2711) Deadlock between concurrent 4LW
 commands that iterate over connections with Netty server
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit
archived-at: Mon, 13 Mar 2017 15:37:49 -0000


     [ https://issues.apache.org/jira/browse/ZOOKEEPER-2711?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Josh Elser reassigned ZOOKEEPER-2711:
-------------------------------------

    Assignee: Josh Elser

> Deadlock between concurrent 4LW commands that iterate over connections with Netty server
> ----------------------------------------------------------------------------------------
>
>                 Key: ZOOKEEPER-2711
>                 URL: https://issues.apache.org/jira/browse/ZOOKEEPER-2711
>             Project: ZooKeeper
>          Issue Type: Bug
>            Reporter: Josh Elser
>            Assignee: Josh Elser
>            Priority: Critical
>
> Observed the following issue in some $dayjob testing environments. Line numbers are a little off compared to master/branch-3.5, but I did confirm the same issue exists there.
> With the NettyServerCnxnFactory, before a request is dispatched, the code synchronizes on the {{NettyServerCnxn}} object. However, with some 4LW commands (like {{stat}}), each {{ServerCnxn}} object is also synchronized to (safely) iterate over the internal contents of the object to generate the necessary debug message. As such, multiple concurrent {{stat}} commands can both lock their own {{NettyServerCnxn}} objects, and then be blocked waiting to lock each others' {{ServerCnxn}} in the {{StatCommand}}, deadlocked.
> {noformat}
> "New I/O worker #55":
> 	at org.apache.zookeeper.server.ServerCnxn.dumpConnectionInfo(ServerCnxn.java:407)
> 	- waiting to lock <0x00000000fabc01b8> (a org.apache.zookeeper.server.NettyServerCnxn)
> 	at org.apache.zookeeper.server.NettyServerCnxn$StatCommand.commandRun(NettyServerCnxn.java:478)
> 	at org.apache.zookeeper.server.NettyServerCnxn$CommandThread.run(NettyServerCnxn.java:311)
> 	at org.apache.zookeeper.server.NettyServerCnxn$CommandThread.start(NettyServerCnxn.java:306)
> 	at org.apache.zookeeper.server.NettyServerCnxn.checkFourLetterWord(NettyServerCnxn.java:677)
> 	at org.apache.zookeeper.server.NettyServerCnxn.receiveMessage(NettyServerCnxn.java:790)
> 	at org.apache.zookeeper.server.NettyServerCnxnFactory$CnxnChannelHandler.processMessage(NettyServerCnxnFactory.java:211)
> 	at org.apache.zookeeper.server.NettyServerCnxnFactory$CnxnChannelHandler.messageReceived(NettyServerCnxnFactory.java:135)
> 	- locked <0x00000000fab68178> (a org.apache.zookeeper.server.NettyServerCnxn)
> 	at org.jboss.netty.channel.SimpleChannelHandler.handleUpstream(SimpleChannelHandler.java:88)
> 	at org.jboss.netty.channel.DefaultChannelPipeline.sendUpstream(DefaultChannelPipeline.java:564)
> 	at org.jboss.netty.channel.DefaultChannelPipeline.sendUpstream(DefaultChannelPipeline.java:559)
> 	at org.jboss.netty.channel.Channels.fireMessageReceived(Channels.java:268)
> 	at org.jboss.netty.channel.Channels.fireMessageReceived(Channels.java:255)
> 	at org.jboss.netty.channel.socket.nio.NioWorker.read(NioWorker.java:88)
> 	at org.jboss.netty.channel.socket.nio.AbstractNioWorker.process(AbstractNioWorker.java:109)
> 	at org.jboss.netty.channel.socket.nio.AbstractNioSelector.run(AbstractNioSelector.java:312)
> 	at org.jboss.netty.channel.socket.nio.AbstractNioWorker.run(AbstractNioWorker.java:90)
> 	at org.jboss.netty.channel.socket.nio.NioWorker.run(NioWorker.java:178)
> 	at org.jboss.netty.util.ThreadRenamingRunnable.run(ThreadRenamingRunnable.java:108)
> 	at org.jboss.netty.util.internal.DeadLockProofWorker$1.run(DeadLockProofWorker.java:42)
> 	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> 	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> 	at java.lang.Thread.run(Thread.java:745)
> "New I/O worker #51":
> 	at org.apache.zookeeper.server.ServerCnxn.dumpConnectionInfo(ServerCnxn.java:407)
> 	- waiting to lock <0x00000000fab68178> (a org.apache.zookeeper.server.NettyServerCnxn)
> 	at org.apache.zookeeper.server.NettyServerCnxn$StatCommand.commandRun(NettyServerCnxn.java:478)
> 	at org.apache.zookeeper.server.NettyServerCnxn$CommandThread.run(NettyServerCnxn.java:311)
> 	at org.apache.zookeeper.server.NettyServerCnxn$CommandThread.start(NettyServerCnxn.java:306)
> 	at org.apache.zookeeper.server.NettyServerCnxn.checkFourLetterWord(NettyServerCnxn.java:677)
> 	at org.apache.zookeeper.server.NettyServerCnxn.receiveMessage(NettyServerCnxn.java:790)
> 	at org.apache.zookeeper.server.NettyServerCnxnFactory$CnxnChannelHandler.processMessage(NettyServerCnxnFactory.java:211)
> 	at org.apache.zookeeper.server.NettyServerCnxnFactory$CnxnChannelHandler.messageReceived(NettyServerCnxnFactory.java:135)
> 	- locked <0x00000000fabc01b8> (a org.apache.zookeeper.server.NettyServerCnxn)
> 	at org.jboss.netty.channel.SimpleChannelHandler.handleUpstream(SimpleChannelHandler.java:88)
> 	at org.jboss.netty.channel.DefaultChannelPipeline.sendUpstream(DefaultChannelPipeline.java:564)
> 	at org.jboss.netty.channel.DefaultChannelPipeline.sendUpstream(DefaultChannelPipeline.java:559)
> 	at org.jboss.netty.channel.Channels.fireMessageReceived(Channels.java:268)
> 	at org.jboss.netty.channel.Channels.fireMessageReceived(Channels.java:255)
> 	at org.jboss.netty.channel.socket.nio.NioWorker.read(NioWorker.java:88)
> 	at org.jboss.netty.channel.socket.nio.AbstractNioWorker.process(AbstractNioWorker.java:109)
> 	at org.jboss.netty.channel.socket.nio.AbstractNioSelector.run(AbstractNioSelector.java:312)
> 	at org.jboss.netty.channel.socket.nio.AbstractNioWorker.run(AbstractNioWorker.java:90)
> 	at org.jboss.netty.channel.socket.nio.NioWorker.run(NioWorker.java:178)
> 	at org.jboss.netty.util.ThreadRenamingRunnable.run(ThreadRenamingRunnable.java:108)
> 	at org.jboss.netty.util.internal.DeadLockProofWorker$1.run(DeadLockProofWorker.java:42)
> 	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> 	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> 	at java.lang.Thread.run(Thread.java:745)
> {noformat}
> It would appear that the synchronization on the {{NettyServerCnxn}} in {{NettyServerCnxnFactory}} is to blame (and I can see why it was done originally). I think we can just use a different Object (and monitor) to provide mutual exclusion at Netty layer (and avoid synchronization issues at the "application" layer).


--
This message was sent by Atlassian JIRA
(v6.3.15#6346)