HDP on Linux – Installation Forum

Manual Installation Ganglia

  • #49345
    Gwenael Le Barzic

    Hello !

    I’m trying to install Ganglia on the following architecture :
    – 1 master
    – 3 slaves

    I already installed Hadoop and some other components thanks to Ambari But I didn’t install Ganglia at this time.

    So, in order to install Ganglia, I downloaded the companion tar.gz file, and followed all the manual installation guide for Ganglia :

    At the end of the installation guide, there are some steps to validate the installation. On the Ganglia master node, which is also my master node for Hadoop, I tried the following command :

    [root@master ~]# /etc/init.d/hdp-gmetad start
    Starting hdp-gmetad...
    Failed to create base directory '-B': Permission denied
    chgrp: impossible d'accéder à « /var/run/ganglia/hdp/rrdcached.sock »: Auc
    un fichier ou dossier de ce type
    chgrp: impossible d'accéder à « /var/run/ganglia/hdp/rrdcached.limited.sockÂ
    »: Aucun fichier ou dossier de ce type
    Failed to start /usr/bin/rrdcached
    Not starting /usr/sbin/gmetad because starting /usr/bin/rrdcached failed.

    I investigated this problem and found that the problem was coming from a variable which is not set. This is why there is an error in the creation of the base directory.

    I create a second topic for this problem, because my previous one was more concerning a problem in my own repo folder.

    -m 664 -l unix:${RRDCACHED_ALL_ACCESS_UNIX_SOCKET} \

    Which gives this once I echoed it after the run of the script :
    su nobody -c /usr/bin/rrdcached -p /var/run/ganglia/hdp/rrdcached.pid
    -m 664 -l unix:/var/run/ganglia/hdp/rrdcached.sock
    -m 777 -P FLUSH,STATS,HELP -l unix:/var/run/ganglia/hdp/rrdcached.limited.sock
    -b -B

    Seems like ${RRDCACHED_BASE_DIR} is not set.

    Do you have any idea what I can test, where I can investigate to understand what is going on ?

    Best regards.


to create new topics or reply. | New User Registration

  • Author
  • #49346
    Gwenael Le Barzic

    Just as a note, on the three slaves, the start of /etc/init.d/hdp-gmond start worked fine.

    In the contrary, on the master, even the command “/etc/init.d/hdp-gmetad start ” failed with the following error message :
    [root@master ~]# /etc/init.d/hdp-gmond start
    Starting hdp-gmond...
    Failed to start /usr/sbin/gmond for cluster HDPHistoryServer

    For your information here is the list of components by host :

    <u>On the master :</u>
    HDFS Client
    History Server
    Hive Client
    Hive Metastore
    MapReduce2 Client
    MySQL Server
    Oozie Client
    Oozie Server
    WebHCat Server
    YARN Client
    ZooKeeper Client
    ZooKeeper Serve

    <u>Slave 1</u>
    ZooKeeper Server

    <u>Slave 2</u>
    ZooKeeper Server

    <u>Slave 3</u>
    ZooKeeper Server

    Best regards.


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.