Currently, the graphite data is grouped in logical groups for Hadoop. The graphite queries which fetch data, do it considering these logical groups and send aggregate data for all hosts in that group to Zabbix. In Zabbix, hosts are the logical groups which encompass individual hosts of same type; for e.g.: hbase_master represents all hbase masters. All this was designed to work with JMX data for hadoop components.
Now, if one wants to monitor and alarm on data collected by other means (say diamond) and wishes to trigger for individual hosts, the current framework does not work. Changes need to be made so arbitrary hosts can be monitored using graphite and Zabbix.
Currently, the graphite data is grouped in logical groups for Hadoop. The graphite queries which fetch data, do it considering these logical groups and send aggregate data for all hosts in that group to Zabbix. In Zabbix, hosts are the logical groups which encompass individual hosts of same type; for e.g.: hbase_master represents all hbase masters. All this was designed to work with JMX data for hadoop components.
Now, if one wants to monitor and alarm on data collected by other means (say diamond) and wishes to trigger for individual hosts, the current framework does not work. Changes need to be made so arbitrary hosts can be monitored using graphite and Zabbix.