# Alibaba Arthas — Java Application Diagnostic Tool

> A guide to Arthas, the open-source Java diagnostic tool that lets you troubleshoot production issues in real time without restarting or modifying code.

## Install

Save in your project root:

# Alibaba Arthas — Java Application Diagnostic Tool

## Quick Use
```bash
# One-line install and attach to a running JVM
curl -O https://arthas.aliyun.com/arthas-boot.jar
java -jar arthas-boot.jar
# Inside Arthas shell:
dashboard          # real-time JVM overview
thread -n 3        # top 3 busiest threads
watch com.example.MyService myMethod returnObj
```

## Introduction
Arthas is an open-source Java diagnostic tool developed by Alibaba. It allows developers to troubleshoot production JVM applications in real time — inspecting class loading, decompiling code, tracing method calls, and profiling performance — all without restarting the application or adding logging.

## What Arthas Does
- Attaches to any running JVM process and opens an interactive diagnostic shell
- Decompiles loaded classes on the fly to verify deployed code matches expectations
- Traces method invocations with timing, arguments, return values, and exceptions
- Monitors JVM metrics including threads, memory, GC, and CPU in a live dashboard
- Allows hot-swapping of class bytecode for emergency production fixes

## Architecture Overview
Arthas uses a Java Agent that attaches to the target JVM via the Attach API. Once connected, it instruments bytecode using ASM to intercept method calls without modifying source. The agent runs a Netty-based server that accepts commands from the Arthas CLI client or a web console, executes them in the target JVM context, and returns results over the tunnel.

## Self-Hosting & Configuration
- Download arthas-boot.jar or install via package managers (brew, sdkman)
- Run java -jar arthas-boot.jar to list JVM processes and select one to attach
- Access the web console at http://localhost:3658 for browser-based diagnostics
- Configure tunnel server for remote diagnosis of distributed applications
- Integrate with Spring Boot Starter for embedded diagnostic endpoints

## Key Features
- Zero-intrusion diagnostics: no code changes, restarts, or additional JVM flags needed
- Real-time class decompilation to verify what bytecode is actually running
- OGNL expression support for evaluating arbitrary expressions in the target JVM
- Flame graph profiling via async-profiler integration for CPU and allocation analysis
- Web console and HTTP API for remote troubleshooting across environments

## Comparison with Similar Tools
- **JVisualVM** — GUI-based JVM monitor; Arthas provides CLI-first deep diagnostics without requiring a display server
- **JProfiler** — Commercial profiler with rich UI; Arthas is free, lighter, and designed for production attachment
- **BTrace** — Bytecode tracing tool; Arthas offers a more user-friendly command interface and broader feature set
- **async-profiler** — CPU/allocation profiler; Arthas wraps it and adds method tracing, class inspection, and OGNL evaluation
- **jcmd/jstack** — JDK built-in tools; Arthas provides an interactive shell combining all their capabilities plus method-level tracing

## FAQ
**Q: Is it safe to attach Arthas to a production JVM?**
A: Yes. Arthas instruments only the methods you explicitly target, and the overhead is negligible for typical diagnostic commands. Detach when done to remove all instrumentation.

**Q: Does Arthas work with GraalVM native images?**
A: Arthas requires a standard HotSpot or OpenJ9 JVM with the Attach API. GraalVM native images do not support dynamic agent attachment.

**Q: Can I use Arthas in containers?**
A: Yes. Mount arthas-boot.jar into the container or include it in the image. Use the tunnel server for Kubernetes pods that lack direct shell access.

**Q: What Java versions does Arthas support?**
A: Arthas supports Java 6 through Java 21+, covering both LTS and recent non-LTS releases.

## Sources
- https://github.com/alibaba/arthas
- https://arthas.aliyun.com/en/

---
Source: https://tokrepo.com/en/workflows/ade75a56-3d7c-11f1-9bc6-00163e2b0d79
Author: AI Open Source