How To Build A Multi-measure Container To Measure The Health Of A Distributed Application
Application delivery can be dependent on more than one physical machine, and can involve for example, a database on one machine, and a web server on another machine. Classic network management systems tend to treat these elements separately, which drives a demand for event correlation to accomplish forensic analysis of a failure involving multiple elements. With pressure coming from management to assess the health of applications and report on the availability and performance, it is useful to apply a grouping mechanism that allows the monitoring system to apply service checks on a variety of disparate machines and their resident applications , as related to an overall application delivery. This technique can give early warnings on the deterioration of key distributed elements necessary to run the application, and reduce the pressure for after-the-fact analysis.
Nagios®* is flexible in this regard, and as such can be configured to show a virtual host that has service checks assigned to distributed elements.
The following steps will show you how to set up a virtual host, and assign distributed services to it.
- Set up a new host in Nagios for the service aggregation. Choose a name that reflects the application, say E-Commerce for example. Give your new virtual host the home address of the monitoring server, 127.0.0.1.
- Copy the commands for the necessary parameters so you can rename and edit.
- If you are using active polling then edit the commands to hard-code the host address instead of using the Nagios variable $HOSTNAME$. Save under a different name to distinguish them from the originals. Then clone the service usually associated with the original un-edited command, rename it, and assign the new check command with the hard-coded host address. Then select your new host in the Host configuration menu and add the new service.
- If you use passive checks kicked off by a cron job on the target hosts, just assign them to your virtual host.
The result of this organizational structure is the capability to display the host, named for the application comprised within, and its attendant service checks in a single view via Status Viewer, or within a dashboard.
* Nagios is a registered trademark of Nagios Enterprises
- If you are using active polling then edit the commands to hard-code the host address instead of using the Nagios variable $HOSTNAME$. Save under a different name to distinguish them from the originals. Then clone the service usually associated with the original un-edited command, rename it, and assign the new check command with the hard-coded host address. Then select your new host in the Host configuration menu and add the new service.
