[jira] [Created] (ZOOKEEPER-4722) Reproducible SessionTimeoutException for ZK3.5.7 and higher

["Olivier Tavard (Jira)" <ji...@apache.org>]

Olivier Tavard created ZOOKEEPER-4722:
-----------------------------------------

             Summary: Reproducible SessionTimeoutException for ZK3.5.7 and higher
                 Key: ZOOKEEPER-4722
                 URL: https://issues.apache.org/jira/browse/ZOOKEEPER-4722
             Project: ZooKeeper
          Issue Type: Bug
    Affects Versions: 3.5.7
            Reporter: Olivier Tavard


Hello Zookeeper team,

I am contributing to the Apache ManifoldCF project, which is heavily relying on Zookeeper, and I encountered a reproducible error for which we – the ManifoldCF team - need your help:

As part of its output connectors regular updates (ManifoldCF is a data crawler that can send analysed documents to designated destination applications), we are upgrading the Apache Solr output connector so that it is compatible with Solr 9.2. This version of Solr uses Apache Zookeeper 3.8.0, but when using the same Zookeeper version on the Apache ManifoldCF side, we experienced reproducible tests failures on the Solr Output Connector of ManifoldCF when building the project.

We were able to pass tests with ZK up to version 3.5.6 included, but all of the more recent versions fail [4].

To ease your evaluation of the problem, we have prepared a fork of Apache ManifoldCF where you can work and reproduce the issue [1]. The branch to be used is CONNECTORS-1740, and the build procedure is as follows (the issue occurs during the testings phase):

git checkout CONNECTORS-1740
 
Now, in order to build with ant (exclusively with Java 11 otherwise it will not work)
First, Fetch the dependencies : 
ant clean make-core-deps
 
Next, start the build itself: 
ant clean build
 
You will get a success message for both commands.
 
Then start the MCF tests (this is what will trigger the error [4]):
ant run-tests-framework

Thanks in advance for your help.

Best regards,

Olivier
 

[1] https://gitlab.datafari.com/sandboxespublic/manifoldcf
You can find here the email thread about this problem on the Apache ManifoldCF developer Mailing List [2], as well as the jira issue dedicated to it [3].

[2] https://lists.apache.org/thread/3zx7c5nm23j6wfb0okf5xb01t9jb6lph

[3] https://issues.apache.org/jira/browse/CONNECTORS-1740

[4] Error logs we get:

[junit] org.apache.zookeeper.ClientCnxn$SessionTimeoutException: Client session timed out, have not heard from server in 2007ms for session id 0x0
[junit]     at org.apache.zookeeper.ClientCnxn$SendThread.run(ClientCnxn.java:1250)
[junit] [reader] INFO org.apache.zookeeper.ZooKeeper - Session: 0x0 closed
[junit] [reader-EventThread] INFO org.apache.zookeeper.ClientCnxn - EventThread shut down for session: 0x0
[junit] [reader] INFO org.apache.zookeeper.ZooKeeper - Initiating client connection, connectString=localhost:8348 sessionTimeout=2000 watcher=org.apache.manifoldcf.core.lockmanager.ZooKeeperConnection$ZooKeeperWatcher@45c66a27
[junit] [reader] INFO org.apache.zookeeper.ClientCnxnSocket - jute.maxbuffer value is 1048575 Bytes
[junit] [reader] INFO org.apache.zookeeper.ClientCnxn - zookeeper.request.timeout value is 0. feature enabled=false
[junit] [reader-SendThread(localhost:8348)] INFO org.apache.zookeeper.ClientCnxn - Opening socket connection to server localhost/127.0.0.1:8348.
[junit] [reader-SendThread(localhost:8348)] INFO org.apache.zookeeper.ClientCnxn - SASL config status: Will not attempt to authenticate using SASL (unknown error)
[junit] [reader-SendThread(localhost:8348)] WARN org.apache.zookeeper.ClientCnxn - Client session timed out, have not heard from server in 2008ms for session id 0x0
[junit] [reader-SendThread(localhost:8348)] WARN org.apache.zookeeper.ClientCnxn - An exception was thrown while closing send thread for session 0x0.
[junit] org.apache.zookeeper.ClientCnxn$SessionTimeoutException: Client session timed out, have not heard from server in 2008ms for session id 0x0



--
This message was sent by Atlassian Jira
(v8.20.10#820010)