Deploying Pinpoint for Distributed Tracing of Dubbo Services

When using Dubbo for service‑oriented architecture, it is often hard to determine which client caused an exception; a distributed tracing system such as Pinpoint can quickly locate the root cause.

What is Pinpoint – Pinpoint is an open‑source APM (Application Performance Management) tool for large‑scale Java distributed systems, inspired by Google Dapper. It visualizes module interactions, provides server maps, real‑time thread charts, request/response scatter charts, call stacks, and inspectors to analyze system topology and performance.

Supported Modules include JDK 6+, various servlet containers (Tomcat, Jetty, JBoss, etc.), Spring/Spring Boot, HTTP clients, Thrift, Dubbo, messaging systems (ActiveMQ, RabbitMQ), databases (MySQL, Oracle, PostgreSQL, etc.), caches (Redis, Cassandra), and logging frameworks.

Deployment Steps

Prepare a basic environment (Hadoop, HBase, Zookeeper) and ensure firewalls are disabled if necessary.

Download Pinpoint binaries from GitHub: wget https://github.com/naver/pinpoint/releases/download/1.7.3/pinpoint-agent-1.7.3.tar.gz wget https://github.com/naver/pinpoint/releases/download/1.7.3/pinpoint-collector-1.7.3.war wget https://github.com/naver/pinpoint/releases/download/1.7.3/pinpoint-web-1.7.3.war

Set up Zookeeper, HBase (single node), and JDK 1.8 (or OpenJDK).

In HBase, run ./hbase shell hbase-create.hbase to create required tables.

Prepare a Tomcat instance for deployment.

Deploy the two WAR files: rename pinpoint-web-1.7.3.war to ROOT.war and pinpoint-collector-1.7.3.war to collector.war , then place them in Tomcat’s webapps directory and start Tomcat.

Verify Tomcat logs for connection errors (e.g., port 2181).

Agent Installation

Extract pinpoint-agent-1.7.3.tar.gz and edit pinpoint.config to set profiler.collector.ip to the collector’s host.

Deploy the agent on every application server; Pinpoint adds only about 3% overhead.

Prepare a Dubbo Sample Project

Clone the sample repository: https://github.com/souyunku/spring-boot-examples/tree/master/spring-boot-dubbo

Update application.properties to point zookeeper.connect to your Zookeeper address.

Build the project with mvn clean package .

Configure Startup Parameters

For Tomcat, edit bin/catalina.sh and add:

CATALINA_OPTS="$CATALINA_OPTS -javaagent:$AGENT_PATH/pinpoint-bootstrap-1.7.3.jar"
CATALINA_OPTS="$CATALINA_OPTS -Dpinpoint.agentId=tomcat1"
CATALINA_OPTS="$CATALINA_OPTS -Dpinpoint.applicationName=webcontroller"

For SpringBoot, start the JVM with:

java -javaagent:/opt/pinpoint-bootstrap-1.7.3.jar -Dpinpoint.agentId=dubbo-provider-1 -Dpinpoint.applicationName=dubbo-provider -jar dubbo-provider-1.0-SNAPSHOT.jar

java -javaagent:/opt/pinpoint-bootstrap-1.7.3.jar -Dpinpoint.agentId=dubbo-consumer-1 -Dpinpoint.applicationName=dubbo-consumer -jar dubbo-consumer-1.0-SNAPSHOT.jar

After launching the services, log into Pinpoint’s web UI to view cluster topology, request traces, and performance metrics such as CPU, memory, GC, TPS, and JVM parameters.

Testing

Send a request to the consumer, e.g., http://localhost:8080/sayHello?name=souyunku , and observe the trace details, server map, and bottleneck locations in the Pinpoint UI.

Several screenshots illustrate the Pinpoint dashboard, request details, and system metrics.