Skip to content
WebLogic Monitoring Exporter exposes metrics and monitoring data through REST APIs for consumption by other tools (e.g. Prometheus)
Branch: master
Clone or download
Latest commit 8f80526 Jul 15, 2019
Type Name Latest commit message Commit time
Failed to load latest commit information.
config_coordinator update copyright info Jun 26, 2019
samples review and edit Jul 16, 2019
src update copyright info Jun 26, 2019
webapp update copyright info Jun 26, 2019
.gitignore ignore target under samples Aug 10, 2017
.travis.yml Create .travis.yml Dec 4, 2017 update copyright info Jun 26, 2019
LICENSE update copyright info Jun 26, 2019 update copyright info Jun 26, 2019
THIRD_PARTY_LICENSES.txt Add third-party license file Jun 26, 2019
pom.xml Update pom.xml Jun 26, 2019

WebLogic Monitoring Exporter

Build Status

The WebLogic Monitoring Exporter uses the WLS RESTful Management API to scrape runtime information and then exports Prometheus-compatible metrics. It is deployed as a web application in a WebLogic Server (WLS) instance, version 12.2.1 or later, typically, in the instance from which you want to get metrics.

Building and configuring

Each release contains a script. Download it and then run

bash <some-config-file>

to download the release and install the specified default configuration.

Building from source

Use mvn install to build the exporter servlets.

Then cd webapp & mvn package to build the web application wls-exporter.war, which can then be deployed to WLS.

Adding -Dconfiguration=<some-config-file> will create a web application with the specified configuration as its default.


Here is an example yaml configuration:

  url: http://coordinator:8999/
  interval: 5
metricsNameSnakeCase: true
domainQualifier: true
restPort: 7001
- key: name
  keyName: server
    key: name
    keyName: app
      type: WebAppComponentRuntime
      prefix: webapp_config_
      key: name
      values: [deploymentState, contextRoot, sourceInfo, openSessionsHighCount]
        prefix: weblogic_servlet_
        key: servletName
        values: invocationTotalCount

Note that there are two parts to the configuration. The optional top portion defines general processing rules.

Name Description
query_sync Optional. Configuration for a service which coordinates updates to the query configuration.
query_sync.url The URL of the service. Required if this section is present.
query_sync.interval The interval, in seconds, at which the service will be queried. Defaults to 10.
metricsNameSnakeCase If true, metrics names will be converted to snake case. Defaults to false.
domainQualifier If true, the domain name will be included as a qualifier for all metrics. Defaults to false.
restPort Optional. Overrides the port on which the exporter should contact the REST API. Needed when behind a load balancer.

The query field is more complex. Each query consists of a hierarchy of the MBeans, starting relative to ServerRuntimes. Within each section, there are a number of options:

Name Description
key The name of the attribute to use as a key for qualifiers in the output.
keyName The name to use for the key in the qualifier; defaults to the name of the attribute.
prefix A prefix to use for all the metrics gathered from the current level.
values The attributes for which metrics are to be output. If not specified and a prefix is defined, all values on the MBean will be selected.
type A filter for subtypes. If specified, only those objects whose type attribute matches will be collected.

Note that all fields other than the above, will be interpreted as collections of values.

In the above example, the presumed underlying data structure is:

+---------------+   applicationRuntimes     
| ServerRuntime |-----------+                 
+---------------+           |
            +--------------------+                       +------------------+
            | ApplicationRuntime |   componentRuntimes   | ComponentRuntime |
            |--------------------| --------------------> |------------------|
            | + name             |                       | + type           |
            +--------------------|                       +------------------+
                                                                  / \
     +------------------------+                    +-------------------------+
     |        Servlet         |       servlets     | WebAppComponentRuntime  |
     |------------------------| <----------------- |-------------------------|
     | + servletName          |                    | + name                  |
     | + invocationTotalCount |                    | + contextRoot           |
     +------------------------+                    | + deploymentState       |
                                                   | + sourceInfo            |
                                                   | + openSessionsHighCount |

The above configuration will then produce metrics such as:


Note that no help or type text is currently produced, as the REST API has no access to the underlying MBean info.


The exporter produces metrics for monitoring its own performance:

  • wls_scrape_mbeans_count_total reports the number of metrics scraped.
  • wls_scrape_duration_seconds reports the time required to do the scrape.
  • wls_scrape_cpu_seconds reports the CPU time used during the scrape.

Access to the REST API

The exporter must be able to contact the REST API of the WLS instance on which it is deployed. It does so by using the primary host address of its server and the port on which the request for metrics was made. Usually, that will work; however, if the metrics request is made via a load balancer or Kubernetes NodePort service, the port to which the original request was made might not be the same as the instance's HTTP port. In such a case, the configuration should include the restPort configuration to tell the exporter which port to use.


  • Deployment-based sample: This sample helps you set up and run Prometheus and Grafana with the WebLogic Monitoring Exporter. Deployment YAML files are provided to install Prometheus and Grafana.

  • Chart-based sample: This is an end-to-end sample that shows you the steps to set up monitoring for WebLogic domains using Prometheus and Grafana. Prometheus and Grafana are installed with Helm charts.


Copyright (c) 2017, 2019, Oracle Corporation and/or its affiliates. All rights reserved.

You can’t perform that action at this time.