Monitoring Tomcat Servers

Using the specialized monitoring model that eG Enterprise offers for the Tomcat server (see Figure 1), one can extensively monitor every layer of the Tomcat server, automatically correlate performance across the layers, and accurately locate the problem source.

Note:

The eG agent can monitor a Tomcat server, only if the Tomcat server executes on JDK 1.5 or above.

Figure 1 : The layer model of the Tomcat server

Each layer depicted by Figure 1 performs a number of tests on the Tomcat server and evaluates the critical aspects of Tomcat performance such as thread usage, memory management, processing capabilities of the servlets, efficiency of the JSP engine, etc.

The eG agent begins monitoring the Tomcat Server once the pre-requisites discussed in the Pre-requsites for monitoring the Tomcat Servers are fulfilled. The statistics so collected enable administrators find quick and accurate answers to persistent performance-related queries, such as the following:

Availability monitoring
  • Is the Tomcat server available?
  • How quickly does it respond to requests?
Thread pool usage monitoring
  • Have adequate threads been allocated to the pool?
  • Are too many threads idle?
Servlet monitoring
  • Is the server taking too long to process servlets? Which servlet is contributing to the delay?
Jasper monitoring
  • How frequently do reloads occur?
  • Are the JSPs been updated often?
Connector monitoring
  • How quickly are connectors processing requests? Is there a delay in request processing? If so, which connector is responsible for the delay?
  • How much traffic is generated by a connector?
  • Have any errors occurred during request processing?
JVM monitoring
  • Has the JVM been up and running continuously?
  • How many classes have been loaded/unloaded by the JVM?
  • Is there a garbage collection bottleneck?
  • Does the JVM have sufficient free memory?
Session manager monitoring
  • Is the server workload high? How many sessions are currently active on the server?
  • Were too many sessions rejected?
Cache monitoring
  • Are cache hits optimal?
  • Are disk accesses low?