Server IPC Version 9

to create new topics or reply. | New User Registration


This topic contains 5 replies, has 5 voices, and was last updated by  Robert Molina 10 months, 3 weeks ago.

  • Creator
  • #43904



    I have recently used HDP 2.0 to set up my master and slaves (centOS).
    I’m writing a Java program to be able to interact with the files in the HDFS.
    faced this error and I’m not sure what the problem is:

    org.apache.hadoop.ipc.RemoteException: Server IPC version 9 cannot communicate with client version 4


Viewing 5 replies - 1 through 5 (of 5 total)

You must be to reply to this topic. | Create Account

  • Author
  • #53618

    Robert Molina

    Hi Umer,
    Did you rebuild your client app using the matching hadoop libs with the cluster you are using?



    Umer Baig

    *** Error 7827 Java SQL Exception SQLSTATE 39001: Invalid SQL state ([TD-
    SQLH]: Error serializing hcatalog object: Server IPC version 9 cannot communi
    cate with client version 4).

    I have Hadoop and trying to connect with sqlh teradata.



    Hello Chris and Xiandong,

    Thanks for the information and the help.
    I have updated the referenced libraries in my Java program from the Hadoop-core-1.x.jar to the Hadoop-core-2.x.jar and it worked.
    I guess I forgot to update the referenced libraries while changing from 1.x to 2.x.



    Chris Nauroth

    Hello Gerald and Xiandong,

    Between the 1.x line and the 2.x line, there were multiple changes made to the Hadoop RPC interfaces, and these changes are not backwards-compatible. The error message indicates that your application tried to use an older client (likely from the 1.x line) to communicate with a 2.x cluster.

    To fix the problem, you’ll need to link your application with a 2.x Hadoop client. For example, if your project uses Maven for its build, then you can set your Hadoop dependency version to 2.2.0. If you typically run your application from a cluster node using the “hadoop jar” command, then it’s also best to specify provided for Hadoop dependencies. This way, the Maven build will not bundle specific Hadoop jar versions into your build artifacts. Instead, “hadoop jar” will provide its currently deployed version for you.

    Are you in a situation where you need to support an application that can run against either a 1.x cluster or a 2.x cluster? Downstream projects like HBase and Hadoop LZO face this problem, and they solve it by defining separate Maven profiles for linking against 1.x vs. linking against 2.x. If you need to do something similar, then I recommend looking at the pom.xml from Hadoop LZO. That’s probably the simplest example.

    Hope this helps,


    Xiandong Su

    Gerald, have you resolved this yet? I am running into the exact same error. Some searching online indicating it is some serialization version compatibility issue.

Viewing 5 replies - 1 through 5 (of 5 total)
Hortonworks Data Platform
The Hortonworks Data Platform is a 100% open source distribution of Apache Hadoop that is truly enterprise grade having been built, tested and hardened with enterprise rigor.
Get started with Sandbox
Hortonworks Sandbox is a self-contained virtual machine with Apache Hadoop pre-configured alongside a set of hands-on, step-by-step Hadoop tutorials.
Modern Data Architecture
Tackle the challenges of big data. Hadoop integrates with existing EDW, RDBMS and MPP systems to deliver lower cost, higher capacity infrastructure.