Configuring Metrics Properties

The <APIM_HOME>/repository/conf/metrics.properties file specifies properties that correspond to the gauges in the Metrics Dashboard. See the topic on using JVM metrics for details on using the metrics dashboard for JVM metrics. The level defined for a property in this file determines the extent to which the relevant gauge in the dashboard should be updated with information. The different levels that can be defined for properties are as follows:

Level	Description
Off	Designates no informational events.
Info	Designates informational metric events that highlight the progress of the application at coarse-grained level.
Debug	Designates fine-grained informational events that are most useful to debug an application.
Trace	Designates finer-grained informational events than the DEBUG.
All	Designates all the informational events.

If no specific level is configured for a property in the metrics.properties file, the metrics root level applies. The root level is defined as shown in the following example in the metrics.properties file.

metrics.rootLevel=OFF

If you want to change the current root level, you can also use the following command.

-Dmetrics.rootLevel=INFO

The levels in metrics.properties file can be configured to any hierarchy. However, if the level defined for an individual property is different to the level defined for its parent in the hierarchy, the level defined for the individual property overrules that of the parent. For example, if we have metric.level.jvm.memory=INFO in the <APIM_HOME>/repository/conf/metrics.properties file, all metrics under jvm.memory have INFO as the configured level. However, if you have metric.level.jvm.memory.heap=TRACE, the TRACE level would apply for the metric.level.jvm.memory.heap property even though it is a child property of jvm.memory.

The properties that are included in this file by default are as follows:

JVM's direct and mapped buffer pools

Class loading

Property	Default Level	Description
metric.level.jvm.class-loading	INFO	The gauge showing the number of classes currently loaded for the JVM.

GC

Property	Default Level	Description
metric.level.jvm.gc	DEBUG	The gauge for showing garbage collection and memory usage. Monitoring this allows you to identify memory leaks and memory thrash, which have a negative impact on performance.

Memory

Property	Default Level	Description
metric.level.jvm.memory	INFO	The gauge for showing the used and committed memory in WSO2 API Manager.
metric.level.jvm.memory.heap	INFO	The gauge for showing the used and committed heap in WSO2 API Manager.
metric.level.jvm.memory.non-heap	INFO	The gauge for showing the used code cache and used CMS Perm Gen in WSO2 API Manager.
metric.level.jvm.memory.total	INFO	The gauge for showing the total memory currently available for the JVM.
metric.level.jvm.memory.pools	OFF	The gauge for showing the used and available memory for JVM in the memory pool.

Operating system load

Property	Default Level	Description
metric.level.jvm.os	INFO	The gauge for showing the current load imposed by the JVM on the operating system.

Threads

Property	Default Level	Description
metric.level.jvm.threads	OFF	The parent property of all the gauges relating to the JVM thread pool. The metric level defined for this property applies to all the rest of the properties in this table. The metric level set via this property to a child property can be overruled if a different level is set for it.
metric.level.jvm.threads.count	DEBUG	The gauge for showing the number of active and idle threads currently available in the JVM thread pool.
metric.level.jvm.threads.daemon.count	DEBUG	The gauge for showing the number of active daemon threads currently available in the JVM thread pool.
metric.level.jvm.threads.blocked.count	OFF	The gauge for showing the number of threads that are currently blocked in the JVM thread pool.
metric.level.jvm.threads.deadlock.count	OFF	The gauge for showing the number of threads that are currently deadlocked in the JVM thread pool.
metric.level.jvm.threads.new.count	OFF	The gauge for showing the number of new threads generated in the JVM thread pool.
metric.level.jvm.threads.runnable.count	OFF	The gauge for showing the number of runnable threads currently available in the JVM thread pool.
metric.level.jvm.threads.terminated.count	OFF	The gauge for showing the number of threads terminated from the JVM thread pool since you started running the WSO2 API Manager instance.
metric.level.jvm.threads.timed_waiting.count	OFF	The gauge for showing the number of threads in the `Timed_Waiting` state.
metric.level.jvm.threads.waiting.count	OFF	The gauge for showing the number of threads in the `Waiting` state in the JVM thread pool. One or more other threads are required to perform certain actions before these threads can proceed with their actions.

Viewing API specific Handler Metrics

Follow the below steps to view API specific metrics for handlers which are exposed through Dropwizard. These are timer metrics which allows us to see how much time is taken for each handler.

Copy the JMX Service URL displayed in the console at API Manager server startup.

INFO - JMXServerManager JMX Service URL  : service:jmx:rmi://localhost:11111/jndi/rmi://localhost:9999/jmxrmi

Then go to <JDK_HOME>/bin. JDK_HOME is the directory where the JDK(Java Development Kit) is installed.
Find the jconsole executable file in the directory and run it on the command/shell prompt.

You can just type jconsole in the command/shell prompt.
In the prompted window, enter the JMX Service URL copied before as a new connection. Type "admin" in both username and password fields and click "Connect".
Now invoke an API in API Manager.

Handler Metrics will only be visible when you invoke an API.
In the jconsole window opened above, go to the MBeans tab. There you can view the metrics implemented by using Dropwizard Metrics Library which starts with the org.wso2.am prefix.