Skip to content

Latest commit

 

History

History
294 lines (225 loc) · 15.2 KB

README.md

File metadata and controls

294 lines (225 loc) · 15.2 KB

WebLogic Monitoring Exporter

The WebLogic Monitoring Exporter is a Prometheus-compatible exporter of metrics from WebLogic Server (WLS) instances, which it obtains by using the WLS RESTful Management API, available in version 12.2.1 or later. Metrics are selected using a YAML configuration file.

The exporter is available in two forms:

  • A web application that you deploy to the server from which metrics are to be extracted. You may include a configuration file directly in the WAR file, and you may temporarily modify the configuration in a running system by using a web form, either by selecting a replacement configuration or one to append to the current one, with the caveat that if both the original and appended configurations have filters, the update will be rejected. If a coordination configurator is running and configured, that temporary configuration will be sent to all servers configured to use it.

  • A separate process that is run alongside a server instance. You supply the configuration to such a process with a PUT command, as described below. WebLogic Server Kubernetes Operator versions 3.2 and later have special support for the exporter in this form. For more information, see Use the Monitoring Exporter with WebLogic Kubernetes Operator.

Configuration

Here is an example yaml file configuration:

query_sync:
  url: http://coordinator:8999/
  interval: 5
metricsNameSnakeCase: true
domainQualifier: true
restPort: 7001
queries:
- key: name
  keyName: server
  applicationRuntimes:
    key: name
    keyName: app
    componentRuntimes:
      type: WebAppComponentRuntime
      prefix: webapp_config_
      key: name
      includedKeyValues: "abc_.*"
      excludedKeyValues: "abc_12.*"
      values: [deploymentState, contextRoot, sourceInfo, openSessionsHighCount]
      stringValues:
        status: [deployed, undeployed]
      servlets:
        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, used in the web application only. 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, used in the web application only. Overrides the port on which the exporter should contact the REST API. Needed if the exporter cannot find the REST API. The most common case is running on a system with the administration port enabled. In that case, you must specify the administration port in this field and access the exporter by using the SSL port.

Note that if unable to contact the REST API using the inferred host and port, the exporter will try the local host name and, if the REST port is specified, the local port.

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.
includedKeyValues An optional filter. If specified, only entries whose key value matches the specified regular expression will generate metrics.
excludedKeyValues An optional filter. If specified, entries whose key value matches the specified regular expression will NOT generate metrics.
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.
stringValues A map of string-valued metric names to a list of case-insensitive possible values. They will be converted to indexes of that list.

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

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

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

The above configuration will then produce metrics such as:

webapp_config_deployment_state{domain="mydomain",server="myserver",app="myapp",name="aWebApp"}                                                             
webapp_config_open_sessions_high_count{domain="mydomain",server="myserver",app="myapp",name="aWebApp"}
webapp_config_status{domain="mydomain",server="myserver",app="myapp",name="aWebApp"}
weblogic_servlet_invocation_total_count{domain="mydomain",server="myserver",app="myapp",name="aWebApp",servletName="servlet1"}                                                             
weblogic_servlet_invocation_total_count{domain="mydomain",server="myserver",app="myapp",name="aWebApp",servletName="simpleServlet"}                                                             

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

Self-Monitoring

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 using 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.

Use the Monitoring Exporter with WebLogic Kubernetes Operator

If you are running operator-managed WebLogic Server domains in Kubernetes, simply add the monitoringExporter configuration element in the domain resource to enable the Monitoring Exporter. For an example, see the following yaml file configuration:

kind: Domain
metadata:
  name: domain2
  namespace: domain_ns
spec:
  monitoringExporter:
    configuration:
      metricsNameSnakeCase: true
      domainQualifier: true
      queries:
      - key: name
        keyName: server
        applicationRuntimes:
          key: name
          keyName: app
          componentRuntimes:
            type: WebAppComponentRuntime
            prefix: webapp_config_
            key: name
            values: [deploymentState, contextRoot, sourceInfo, openSessionsHighCount]
            servlets:
              prefix: weblogic_servlet_
              key: servletName
              values: invocationTotalCount

To use the Monitoring Exporter with Prometheus, see the directions here.

Samples

  • 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.

Web application

One way to use the exporter is by creating a WAR file with a default configuration and deploying it to a WebLogic Server instance.

Setting the configuration

The web application has a main landing page, which displays the current configuration and allows you to change it, either by uploading a replacement or an addition to the queries specified with the current one. Metrics will then be available from <application-root>/metrics.

Download the release

You can find all the exporter releases on the Releases page.

To download the web application wls-exporter.war file and put your configuration file into the WAR, download the getXXX.sh script, which is provided with each release and also can be downloaded from the Releases page, and then run:

bash getXXX.sh <your-config-file>

Build from source

Use mvn install to build the web application. This will create wls-exporter-<version>, where version is the Maven-assigned version number.

Adding -Dconfiguration=<some-config-file> will insert the specified configuration as its default and remove the version number to simplify deployment to WebLogic Server.

Sidecar

The sidecar is a standalone process that runs the exporter.

Build and run with Maven

There are two ways to build the sidecar implementation. The first is with Maven, using the same mvn install command specified above. Note that this requires JDK21 or later; building the project with JDK8 will skip the sidecar module. The alternative is to build with Docker.

Instead of manually building them, if you want to use pre-built images, then you can pull a pre-created image from the Oracle Container Registry (OCR) or from Google Container Registry (https://ghcr.io/), as follows:

docker pull container-registry.oracle.com/middleware/weblogic-monitoring-exporter:x.x.x
OR
docker pull ghcr.io/oracle/weblogic-monitoring-exporter:x.x.x

After building, run:

java -jar wls-exporter-sidecar/target/wls-exporter-sidecar.jar

This will start the exporter on port 8080, and it will expect to find a local WebLogic Server instance, listening on port 7001.

You can make changes by specifying parameters on the command line:

Setting Default Property
Domain name (use WLS definition) DOMAIN
Exporter port 8080 EXPORTER_PORT
WebLogic host name localhost WLS_HOST
WebLogic port 7001 WLS_PORT
Use https false WLS_SECURE

Configure the exporter

You configure the sidecar by sending a PUT request to the path /configuration. You can do this with curl; that is how the operator does it.

curl -X PUT -i -u myname:mypassword \
    -H "content-type: application/yaml" \
    --data-binary "@<path to yaml>" \
    http://localhost:8080/configuration

Replace myname and mypassword with the credentials expected by WebLogic Server for its REST API, and <path to yaml> with the relative path to the configuration to use.

Access the metrics

After the exporter is configured, a GET to http://localhost:8080/metrics (or whatever port was chosen) will return the current metrics.

Build a Docker image

If you don't want to use a pre-built image from OCR, you can use the following Docker command to build the Monitoring Exporter image.

docker build . -t <image-name>

This will build the project and create a Docker image with the specified name. It is not necessary even to do the Maven build first, as that will happen as part of creating the image. When running behind a firewall, it is necessary to specify a value for the MAVEN_OPTS and https_proxy variables on the command line. For example:

docker build . --build-arg MAVEN_OPTS="-Dhttps.proxyHost=www-proxy -Dhttps.proxyPort=80" \
               --build-arg https_proxy=www-proxy:80 \
               -t <image-name>

This allows Docker to download the dependencies.

Contributing

This project welcomes contributions from the community. Before submitting a pull request, please review our contribution guide.

Security

Please consult the security guide for our responsible security vulnerability disclosure process.

License

Copyright (c) 2019, 2023, Oracle and/or its affiliates.

Released under the Universal Permissive License v1.0 as shown at https://oss.oracle.com/licenses/upl/.