Master RocketMQ Console: From Zero to Full Monitoring in Minutes

RocketMQ provides a dedicated console (RocketMQ Dashboard) that enables real‑time monitoring of topics, brokers, producers, consumers, and message details, helping developers and operators quickly diagnose problems like unknown consumption status, TPS anomalies, and dead‑letter handling.

Installation

Prerequisites: Java and Maven installed, and a running NameServer and Broker.

# 1 Download binary package
wget https://github.com/apache/rocketmq-dashboard/archive/refs/tags/rocketmq-dashboard-2.0.0.tar.gz
# 2 Extract package
tar -zxvf rocketmq-dashboard-2.0.0.tar.gz
cd rocketmq-dashboard-rocketmq-dashboard-2.0.0
# 3 Build JAR (skip tests)
mvn clean package -Dmaven.test.skip=true
# 4 Start the dashboard
nohup java -jar rocketmq-dashboard-2.0.0.jar \
  --server.port=19095 \
  --rocketmq.config.namesrvAddr=127.0.0.1:9876 > dashboard.log 2>&1 &
# 5 View logs and process
tail -fn 10 dashboard.log
ps -aux|grep rocketmq-dashboard
# 6 Stop the process
kill -15 <pid>
# 7 Access the UI
http://localhost:19095

Using the Dashboard

Dashboard Overview

Switch the interface to Simplified Chinese for easier navigation.

Cluster

Shows all broker nodes, production TPS, consumption TPS, and overall message throughput (approximately 15 M TPS).

Topic

Lists all topics with filters; you can send test messages, reset consumption offsets, or skip backlog to quickly consume messages.

Consumer

Displays all consumer groups, their throughput, delayed messages, and client IPs. Client IDs follow the format clientIP@BrokerIP:PORT@unitName and must be unique; otherwise NameServer cannot route messages correctly.

When deploying in Docker, the container’s PID is always 1, causing identical client IPs. If unitName is not set, RocketMQ may generate a unique value based on the JVM start timestamp to avoid ID collisions.

Message

You can view each message’s content, consumption time, status, and even resend it. Filtering options include messageId, messageKey, and topic.

Message tracking types include NOT_ONLINE (consumer not running), CONSUMED (already consumed), NOT_CONSUME_YET (not yet consumed), UNKNOW_EXCEPTION (error), and CONSUMED_BUT_FILTERED (filtered by tag).

Dead‑Letter Messages

After exceeding retry limits, failed messages go to a dead‑letter queue; once the consumer is fixed, you can batch‑re‑deliver these messages.

The RocketMQ Dashboard is not only a visualization tool but also a powerful operational aid that can reduce troubleshooting time by up to 80%.