HDP on Windows – Installation Forum

Install OK, but trouble validating

  • #26848
    John Bunch
    Member

    I need some help troubleshooting this. I can start all services manually on all nodes, but when I run start_remote_hdp_services.cmd, I get the following error:


    D:\hdp\hadoop>start_remote_hdp_services.cmd
    Master nodes: start hadoop1.sludgebucket.com hadoop2.sludgebucket.com
    0 Master nodes successfully started.
    2 Master nodes failed to start.

    PSComputerName Service Message Status
    -------------- ------- ------- ------
    Connecting to re...
    Connecting to re...

    StartStop-HDPservices : Manually start services on Master nodes then retry
    full cluster start. Exiting.
    At D:\hdp\hadoop\manage_remote_hdp_services.ps1:187 char:26
    + if ($mode -eq "start") { StartStop-HDPservices($mode) }
    + ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
    + CategoryInfo : NotSpecified: (:) [Write-Error], WriteErrorExcep
    tion
    + FullyQualifiedErrorId : Microsoft.PowerShell.Commands.WriteErrorExceptio
    n,StartStop-HDPServices

    I also find this entry in the hadoop-datanode-HADOOP5.log file:


    2013-06-03 09:41:28,818 WARN org.apache.hadoop.hdfs.server.datanode.DataNode: java.io.IOException: Call to hadoop1.FloridaKeys.sajes.com/172.16.1.181:8020 failed on local exception: java.io.IOException: An existing connection was forcibly closed by the remote host
    at org.apache.hadoop.ipc.Client.wrapException(Client.java:1107)
    at org.apache.hadoop.ipc.Client.call(Client.java:1075)
    at org.apache.hadoop.ipc.RPC$Invoker.invoke(RPC.java:225)
    at com.sun.proxy.$Proxy5.sendHeartbeat(Unknown Source)
    at org.apache.hadoop.hdfs.server.datanode.DataNode.offerService(DataNode.java:909)
    at org.apache.hadoop.hdfs.server.datanode.DataNode.run(DataNode.java:1463)
    at java.lang.Thread.run(Unknown Source)
    Caused by: java.io.IOException: An existing connection was forcibly closed by the remote host
    at sun.nio.ch.SocketDispatcher.read0(Native Method)
    (leaving out the rest)

    Here is clusterproperties.txt:


    #Log directory
    HDP_LOG_DIR=d:\hadoop\logs

    #Data directory
    HDP_DATA_DIR=d:\hdp\data

    #Hosts
    NAMENODE_HOST=hadoop1.sludgebucket.com
    SECONDARY_NAMENODE_HOST=hadoop2.sludgebucket.sajes.com
    JOBTRACKER_HOST=hadoop1.sludgebucket.com
    HIVE_SERVER_HOST=hadoop2.sludgebucket.com
    OOZIE_SERVER_HOST=hadoop2.sludgebucket.com
    TEMPLETON_HOST=hadoop2.sludgebucket.com
    SLAVE_HOSTS=hadoop3.sludgebucket.com, hadoop4.sludgebucket.com, hadoop5.sludgebucket.com

    #Database host
    DB_FLAVOR=derby
    DB_HOSTNAME=hadoop2.sludgebucket.com

    #Hive properties
    HIVE_DB_NAME=hivedb
    HIVE_DB_USERNAME=hive_user
    HIVE_DB_PASSWORD=Pa$$w0rd

    #Oozie properties
    OOZIE_DB_NAME=ooziedb
    OOZIE_DB_USERNAME=oozie_user
    OOZIE_DB_PASSWORD=Pa$$w0rd

    I’ve double-checked firewall config and eliminated that as a cause.

to create new topics or reply. | New User Registration

  • Author
    Replies
  • #26850
    John Bunch
    Member

    Windows Server 2012 on all nodes, BTW.

    #27269
    tedr
    Moderator

    Hi John,

    A quick question here, I don’t think that it is the cause of your issue, but why is the domain for the secondary name node different from all the rest?

    Thanks,
    Ted.

    #27284
    John Bunch
    Member

    Ted,

    It isn’t. That’s a mistake in the above text. I was trying to obfuscate the actual domain name but missed a line. Please don’t tell anyone!

    Any idea on what’s causing the error?

    #27395
    tedr
    Moderator

    Hi BalckMamba,

    In a search over the web for the error in the log it points to either a firewall issue or a ssl certificate issue. Check that these are not the cause.

    Thanks,
    Ted.

    #33397
    John Bunch
    Member

    Does not appear to be a firewall issue – if I stop Windows Firewall on all nodes I get the same error.

    I’m not sure how to check for an SSL certificate issue. I have not installed an SSL certificate or made any modifications to Apache – nothing is changed from the installation.

    #33602
    Seth Lyubich
    Moderator

    Hi John,

    I saw similar issue in HDP 1.1. Can you please try to start service with using start_local_hdp_service.cmd script instead?

    I also wanted to note that HDP 1.3 is out which you can try now.

    http://hortonworks.com/products/hdp-windows/

    Hope this helps,

    Thanks,
    Seth

    #33858
    John Bunch
    Member

    Update on this issue:

    The above appears to be due to a Powershell security issue when installing HDP on standalone (non-domain member) Windows servers. There may be a step or two missing in the installation instructions in section 5.5. I added each machine to a domain, set Group Policy for the domain (as directed by section 5.6 and the above issue disappeared. Also, I installed version 1.3 and replicated the behavior.

    #53341
    IngEnsi
    Participant

    Hi ,
    I am new to Hadoop , soryy for this question but I’m wondering is the Active Domain Name should be the same for all Machines.
    for example I want to cinfigure a cluster with 3 nodes : Node1 , Node2 , Node3
    should I create a same domain name Hadoop.com for alla hosts so i get Node1.hadoop.com , Node2.hadoop.com and Node3.hadoop.com
    or should I create a different domain name for each Machine ???

    Thanks
    HerArf

    #53375
    IngEnsi
    Participant

    Can anyone answer me ? please help

    #53442
    Ivan Malamen
    Moderator

    Please check if WinRM and remote powershell execution is enabled on all of the nodes. Remopte start script uses WinRM to connect to all of the nodes.

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

Support from the Experts

A HDP Support Subscription connects you experts with deep experience running Apache Hadoop in production, at-scale on the most demanding workloads.

Enterprise Support »

Become HDP Certified

Real world training designed by the core architects of Hadoop. Scenario-based training courses are available in-classroom or online from anywhere in the world

Training »

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.