Chapter 15. Tuning Ambari Performance for large (>2K-node) clusters

For an HDP cluster having more than 2000 nodes, we recommend that you tune performance in the following ways:

  • Calculate and set a larger task cache size on the Ambari Server.

  • Disable Nagios macros for large clusters to reduce the time Nagios spends processing macro definitions.

  • Increase the maximum number of open files allowed on the Ganglia and Nagios hosts.

  1. Calculate an appropriate cache size using the following formula:

    ecCacheSizeValue = 60 * {cluster_size}, where cluster_size is the number of nodes in the cluster.

  2. On the Ambari Server host, in /etc/ambari-server/conf/ambari-properties, add the following property and value:


  3. On the Ambari Server host, disable environment macros for large clusters by making the following changes:

    • For HDP2, make these changes in /var/lib/ambari-server/resources/stacks/HDP/2.0.6/services/NAGIOS/package/templates/nagios.cfg.j2

    • For HDP1, make these changes in /var/lib/ambari-server/resources/stacks/HDP/1.3.2/services/NAGIOS/package/templates/nagios.cfg.j2

  4. On Ganglia and Nagios hosts, in /etc/security/limits.conf replace the ulimit value with a higher number.


    Ganglia runs as user = nobody, by default. Nagios runs as user = nagios, by default.

  5. Restart Ambari Server.

    ambari-server restart
  6. Restart Nagios.

    Using Ambari Web > Services > Nagios > Service Actions, choose Restart All.

loading table of contents...