This article discusses integrating a Spring-Boot 2.x Java app with Prometheus. The app in question needs to already be instrumented with the Prometheus libraries.

As discussed in a previous post,, Petclinic has been instrumented to provide Prometheus metrics at a /manage/prometheus endpoint.

In this article we will run the instrumented Petclinic application as a Docker container. We will then run a Prometheus service (again as a Docker container) retrieve metrics from the Spring-Boot app to display in the Prometheus dashboard.

Read more after the break.

Create a Petclinic Dockerfile

We use the Petclinic sample app that was previously instrumented. The instrumented JAR is present here: v2.0.0-BUILD-SNAPSHOT. Some knowledge of Docker and a working Docker installation is advised.

  1. Create a new “Docker Workspace” directory for dockerizing Petclinic, say dockerpetclinic.
  2. Create the following Dockerfile.

# Base image that contains OpenJDK 8.0
FROM openjdk:8u131-jdk-alpine

# Labels allow for metadata and visibility
LABEL com.poc.maintainer="Vish"
LABEL com.poc.description="Petclinic Application JAR"

# Persistence for the application directory.
VOLUME /app

# Application and configuration
ADD https://github.com/savishy/spring-petclinic/releases/download/v2.0.0-BUILD-SNAPSHOT/spring-petclinic-2.0.0.BUILD-SNAPSHOT.jar /app/petclinic.jar
ADD application.properties /app/application.properties
ENV JAVA_OPTS=""

# Spring app needs to take local application.properties
ENV SPRING_OPTS="--spring.config.location=/app/application.properties"
ENTRYPOINT [ "sh", "-c", "java $JAVA_OPTS -jar /app/petclinic.jar $SPRING_OPTS" ]

Create an application.properties

The above Dockerfile requires an application.properties alongside it. We use the same default application.properties from the Petclinic source with one modification: server.port=9080 to allow Petclinic to listen on a non-default port.

# database init, supports mysql too
database=hsqldb
spring.datasource.schema=classpath*:db/${database}/schema.sql
spring.datasource.data=classpath*:db/${database}/data.sql

# Web
spring.thymeleaf.mode=HTML

# JPA
spring.jpa.hibernate.ddl-auto=none

# Internationalization
spring.messages.basename=messages/messages

# Actuator / Management
management.endpoints.web.base-path=/manage
management.endpoints.web.exposure.include=*

# Logging
logging.level.org.springframework=INFO
# logging.level.org.springframework.web=DEBUG
# logging.level.org.springframework.context.annotation=TRACE

# Maximum time static resources should be cached
spring.resources.cache.cachecontrol.max-age=12h

server.port=9080

The key piece in this is:

management.endpoints.web.base-path=/manage
management.endpoints.web.exposure.include=*

This means all the management endpoints are enabled. This has an impact on security so be prudent about exposing management endpoints.

Create a Petclinic Docker Image

Build an image named petclinic:

$ cd dockerpetclinic
$ docker build -t petclinic .

Sample Output:

...
...
Step 5/9 : ADD https://github.com/savishy/spring-petclinic/releases/download/v2.0.0-BUILD-SNAPSHOT/spring-petclinic-2.0.0.BUILD-SNAPSHOT.jar /app/petclinic.jar
Downloading [==================================================>]   38.8MB/38.8MB
 ---> fc8430d68a7a
...
...
Step 9/9 : ENTRYPOINT [ "sh", "-c", "java $JAVA_OPTS -jar /app/petclinic.jar $SPRING_OPTS" ]
 ---> Running in bc4b76359f7d
...
...
Successfully built e5476165fb4d
Successfully tagged petclinic:latest

Run the Petclinic App!

docker run -p 9080:9080 -d petclinic

Verify you can load the Prometheus metrics. Open http://localhost:9080/manage/prometheus.

$ 
# HELP system_load_average_1m The sum of the number of runnable entities queued to available processors and the number of runnable entities running on the available processors averaged over a period of time
# TYPE system_load_average_1m gauge
system_load_average_1m 0.0732421875
# HELP tomcat_servlet_request_max_seconds
# TYPE tomcat_servlet_request_max_seconds gauge
tomcat_servlet_request_max_seconds{name="default",} 0.0
# HELP tomcat_sessions_expired_total
# TYPE tomcat_sessions_expired_total counter

Create a Prometheus Docker Image

Prometheus already has a Docker Image here. This is the official image, but needs minor customization for our use-case. We need to configure Prometheus to read the metrics from our Petclinic app.

Note: The Prometheus container usually runs in the same server as the application container. Also, Prometheus and the application are launched within the same docker network.

  1. First create a new docker workspace mkdir ~/dockerprometheus
  2. Create a YAML file prometheus.yml:
global:
  scrape_interval: 10s

scrape_configs:
- job_name: 'petclinic-app'
  metrics_path: /manage/prometheus/
  static_configs:
  - targets: ['petclinic:9080']
  1. Next, create a Dockerfile. This will add our YAML file to the Prometheus base image.
FROM prom/prometheus:v2.3.2
COPY prometheus.yml /etc/prometheus/prometheus.yml

Let’s build a Docker Image named prometheus.

cd ~/dockerprometheus
docker build -t prometheus .

Run Docker Containers!

Notice that the prometheus.yml references the targets as petclinic:9080 i.e. it will look for a DNS entry http://prometheus. This is possible only if we create a bridge network and place both our Petclinic and Prometheus containers in it.

First create a bridge network named dev:

docker network create dev

Create a prometheus and a petclinic container from the images created above. Place both containers in the dev network.

docker run -p 9080:9080 -d --net dev --name petclinic petclinic
docker run -d -p 9090:9090 --net dev --name prometheus prometheus

Verify

Look at the Prometheus service dashboard. You should see one target with a state of “UP”.