zookeeper-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Yuval Dori (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (ZOOKEEPER-2172) Cluster crashes when reconfig a new node as a participant
Date Wed, 07 Mar 2018 11:02:00 GMT

    [ https://issues.apache.org/jira/browse/ZOOKEEPER-2172?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16389390#comment-16389390

Yuval Dori commented on ZOOKEEPER-2172:

1. The use case here was adding node to ZK cluster using zookeeper-3.5.jar. It's not the
same use case as for our customers. the first use 5 machines with 3 ZK instances. shutdown
2 machine (one with ZK. so 2 ZK left) and got "java.lang.IllegalStateException: instance must
be started before calling this method". The second customer got this error when deploying
the application.

This is this issue stack trace: 

2015-04-20 12:53:52,230 [myid:1] - ERROR [LearnerHandler-/]
- Unexpected exception causing shutdown while sock still open 
at java.io.DataInputStream.readInt(Unknown Source) 
at org.apache.jute.BinaryInputArchive.readInt(BinaryInputArchive.java:63) 
at org.apache.zookeeper.server.quorum.QuorumPacket.deserialize(QuorumPacket.java:83) 
at org.apache.jute.BinaryInputArchive.readRecord(BinaryInputArchive.java:103) 
at org.apache.zookeeper.server.quorum.LearnerHandler.run(LearnerHandler.java:493) 
2015-04-20 12:53:52,231 [myid:1] - WARN [LearnerHandler-/]
- ******* GOODBYE / ******** 

And this our customers stack trace: 

2018-02-15T09:58:12.094+0100; ERROR; WSOSTSLXWIT01/MANAGER; P3424/T194; [SPACE/LearnerHandler-/];
Unexpected exception causing shutdown while sock still open 
java.net.SocketTimeoutException: Read timed out 
at java.net.SocketInputStream.socketRead0(Native Method) 
at java.net.SocketInputStream.socketRead(SocketInputStream.java:116) 
at java.net.SocketInputStream.read(SocketInputStream.java:171) 
at java.net.SocketInputStream.read(SocketInputStream.java:141) 
at java.io.BufferedInputStream.fill(BufferedInputStream.java:246) 
at java.io.BufferedInputStream.read(BufferedInputStream.java:265) 
at java.io.DataInputStream.readInt(DataInputStream.java:387) 
at org.apache.jute.BinaryInputArchive.readInt(BinaryInputArchive.java:63) 
at org.apache.zookeeper.server.quorum.QuorumPacket.deserialize(QuorumPacket.java:83) 
at org.apache.jute.BinaryInputArchive.readRecord(BinaryInputArchive.java:99) 
at org.apache.zookeeper.server.quorum.LearnerHandler.run(LearnerHandler.java:542) 

As you can see the row lines for BinaryInputArchive.java and LearnerHandler.java are different
but I thought its related to the different versions (3.4.8 vs 3.5). 


The first customer tested it with ZK 3.5.3 and it didn't reproduced!

What is this new feature that was added to 3.5?

I'll be happy to hear whether do you think if it's related or not.




> Cluster crashes when reconfig a new node as a participant
> ---------------------------------------------------------
>                 Key: ZOOKEEPER-2172
>                 URL: https://issues.apache.org/jira/browse/ZOOKEEPER-2172
>             Project: ZooKeeper
>          Issue Type: Bug
>          Components: leaderElection, quorum, server
>    Affects Versions: 3.5.0
>         Environment: Ubuntu 12.04 + java 7
>            Reporter: Ziyou Wang
>            Assignee: Mohammad Arshad
>            Priority: Critical
>             Fix For: 3.5.3, 3.6.0
>         Attachments: ZOOKEEPER-2172-02.patch, ZOOKEEPER-2172-03.patch, ZOOKEEPER-2172-04.patch,
ZOOKEEPER-2172-06.patch, ZOOKEEPER-2172-07.patch, ZOOKEEPER-2172.patch, ZOOKEPER-2172-05.patch,
history.txt, node-1.log, node-2.log, node-3.log, zoo-1.log, zoo-2-1.log, zoo-2-2.log, zoo-2-3.log,
zoo-2.log, zoo-2212-1.log, zoo-2212-2.log, zoo-2212-3.log, zoo-3-1.log, zoo-3-2.log, zoo-3-3.log,
zoo-3.log, zoo-4-1.log, zoo-4-2.log, zoo-4-3.log, zoo.cfg.dynamic.10000005d, zoo.cfg.dynamic.next,
zookeeper-1.log, zookeeper-1.out, zookeeper-2.log, zookeeper-2.out, zookeeper-3.log, zookeeper-3.out
> The operations are quite simple: start three zk servers one by one, then reconfig the
cluster to add the new one as a participant. When I add the  third one, the zk cluster may
enter a weird state and cannot recover.
>       I found “2015-04-20 12:53:48,236 [myid:1] - INFO  [ProcessThread(sid:1 cport:-1)::PrepRequestProcessor@547]
- Incremental reconfig” in node-1 log. So the first node received the reconfig cmd at 12:53:48.
Latter, it logged “2015-04-20  12:53:52,230 [myid:1] - ERROR [LearnerHandler-/]
- Unexpected exception causing shutdown while sock still open” and “2015-04-20 12:53:52,231
[myid:1] - WARN  [LearnerHandler-/] - ******* GOODBYE  /
********”. From then on, the first node and second node rejected all client connections
and the third node didn’t join the cluster as a participant. The whole cluster was done.
>      When the problem happened, all three nodes just used the same dynamic config file
zoo.cfg.dynamic.10000005d which only contained the first two nodes. But there was another
unused dynamic config file in node-1 directory zoo.cfg.dynamic.next  which already contained
three nodes.
>      When I extended the waiting time between starting the third node and reconfiguring
the cluster, the problem didn’t show again. So it should be a race condition problem.

This message was sent by Atlassian JIRA

View raw message