# Scouter APM: Java and host agents, a collector server and an RCP desktop client, with Telegraf and Zipkin as ways to cover everything else > scouter-project/scouter is an open source application performance monitoring tool written mostly in Java. This review covers its agent, server and client split, what each agent can watch, the Telegraf and Zipkin routes, the plugin ecosystem, and the licence field GitHub could not identify. **scouter-project/scouter** — Scouter is an open source APM (Application Performance Management) tool. - Repository: https://github.com/scouter-project/scouter - Stars: 2,178 · Forks: 571 - Language: Java - License: NOASSERTION - Published: 2026-10-16 · Updated: 2026-10-30 · Language: en - Canonical page: https://hysenlabs.com/projects/scouter-project-scouter ## Agents collect, one server stores, and a desktop client shows Scouter splits into three official parts. Agents gather performance information and send it to the server. The server, also called the collector, saves metrics from Scouter agents or from Telegraf and streams the data to the client. The client, called the viewer, is built on RCP, and the README notes it does not support OSX Big Sur. The repository layout follows the split. Top-level folders include scouter.agent.java, scouter.agent.host, scouter.agent.batch, scouter.server, scouter.client, scouter.common, scouter.webapp and scouter.weaver, built together by a root pom.xml, with build_client.sh and build_package.sh as helper scripts. The README describes Scouter as an open source APM like New Relic and AppDynamics, where APM means application performance monitoring or management. That comparison is the author's framing and the README gives no feature table against either product. ## The Java agent targets Tomcat, JBoss and Resin, and the host agent targets Linux, Windows and Unix Two agents are described. The Java agent monitors web applications running on Tomcat, JBoss, Resin and similar servers, and also standalone Java applications. It gathers profiles and performance metrics from the JVM and the web application server. The host agent covers Linux, Windows and Unix, and in the module list it gathers metrics from Linux, Windows and OSX. A third agent, for MariaDB, is listed as to be announced, so it is not available. A folder named scouter.agent.batch also exists in the repository, though the README does not describe it. The agent split decides what you can see. Application-level data, such as service profiles, comes from the Java agent. Machine-level data, such as CPU, comes from the host agent. The README gives no overhead figure for either agent, so measure the cost in your own environment. ## What Scouter shows: users, services and resources, with profiles for methods, SQL and external calls The README groups metrics in three tiers that follow a chain: users use services, and services use resources. Metrics about users are active users, recently used users and today's visitors. Metrics about services are active services, TPS, response time and application profiles, which include method profile, SQL profile and external call profile. Metrics about resources are CPU, memory, network, heap usage and connection pools. The XLog, a scatter chart of transactions, is the signature view. The README links a guide called How to analyze XLog View in the client documentation, and the project's own blog series covers active services and XLog details. Alerts are customizable through plugins, with a guide named Alert Plugin Guide. The viewer is documented in a client screen help page. All of these live in the scouter.document folder in the repository, which is where the documentation is kept rather than on a separate site. ## Telegraf since 2.0.0 and Zipkin since 2.5.0 widen the targets beyond Java Scouter 2.0.0 added Telegraf support. The README lists the targets that come through it: Redis, nginx, Apache httpd, HAProxy, Kafka, MySQL, MongoDB, RabbitMQ, ElasticSearch, Kube, Mesos and others. A Telegraf server feature guide covers the setup. In this mode the server takes metrics from Telegraf as well as from Scouter agents. Version 2.5.0 added a Zipkin storage module. Any Zipkin instrumentation, for C#, Go, Python, JavaScript or PHP, can then show up in an XLog scatter chart. This routes non-Java applications into the same view, with the zipkin-scouter-storage project documenting how. The README lists method, SQL and external call profiles under the Java agent and says only that Zipkin data appears in the XLog chart. Check the zipkin-scouter documentation before assuming parity for a Go or Python service. ## The Web API since 1.8.0 and Weaver since 2.17.0 give programmatic access Two later modules open Scouter to code. The Web API, since 1.8.0, exposes counters, XLogs, profiles and other performance metrics over HTTP, with a Web API Guide in scouter.document/tech. It is the route for building your own dashboard instead of using the RCP client. Weaver, since 2.17.0, goes the other way. It provides the ability to directly control Scouter XLog and profiles at the code level of Java applications, so a developer can shape what gets recorded from inside the application. A Scouter Weaver Guide is linked, and a scouter.weaver folder is in the repository. On the user interface side, the README lists scouter paper as a third-party UI, with a homepage and a showcase video. It is a separate project in scouter-contrib, so its maintenance and compatibility are outside this repository. Match its version to your server before relying on it. ## Alert and counter plugins live in separate repositories Server plugins extend the collector. A sample, scouter-plugin-server-null, prints out the data collected. Alert plugins forward Scouter alerts to email, Telegram, Slack, LINE, DingTalk and Microsoft Teams. A counter plugin, scouter-plugin-server-influxdb, transfers performance data to InfluxDB, a time series database. Most of these sit under scouter-contrib, not under scouter-project. The agent plugin line reads TBD, so the agent side has nothing listed to install. A third-party Pulse type agent, scouter-pulse-library, includes an aws-monitor that gathers metrics for EC2, RDS and ELB from CloudWatch. Because plugins come from separate repositories, each has its own releases and its own last-push date. The README does not state which plugin versions match which Scouter server version. Pull requests to the main repository are accepted only on the develop branch, and a first pull request requires a CLA, per the README. ## The newest release listed is v2.21.3 from February 2026, and the licence reads NOASSERTION Three releases are listed: v2.21.3 on 2026-02-15, v2.21.2 on 2026-01-17 and v2.21.1 on 2025-12-06. The last push was on 2026-10-11, so commits continued after the latest tag. The repository is not archived. The default branch is master, while pull requests target develop. The README has no install commands. It points to the Quick Start guide, the Installation guide and a live demo, all inside scouter.document, and to the latest release page for downloads. You build or download packages through those routes, and the repository supplies build_package.sh and build_client.sh for building yourself. GitHub reports the licence as NOASSERTION, which means it could not map LICENSE.md to a known identifier. Read that file before redistributing or embedding Scouter. Support goes through a Facebook user group and a Google Groups forum, and the README also links a series of blog posts, translated through Google Translate, that cover installation, XLog and alerts. ## Conclusion Use Scouter if you run Java web applications on Tomcat, JBoss or Resin and want transaction profiles, XLog charts and host metrics from a self-hosted collector. Skip it if you want a browser-only product with a hosted option, because the official viewer is a desktop RCP client. Before adopting it, read LICENSE.md and the Quick Start guide in scouter.document, since GitHub reports the licence as unrecognised and the README itself holds no install commands. ## FAQ ### What is Scouter? Scouter is an open source APM, short for application performance monitoring or management. It has Java and host agents, a collector server and an RCP desktop client. ### What can Scouter monitor? The Java agent covers web applications on Tomcat, JBoss and Resin and standalone Java apps, and the host agent covers Linux, Windows and Unix. Telegraf adds Redis, nginx, Kafka, MySQL and more, and Zipkin adds other languages. ### Where do I find the Scouter install steps? The README has no install commands. It links the Quick Start guide and Installation guide under scouter.document, plus a live demo and a releases page for downloads. ## Sources - [Issues](https://github.com/scouter-project/scouter/issues) - [README](https://github.com/scouter-project/scouter/blob/master/README.md) - [Releases](https://github.com/scouter-project/scouter/releases) - [scouter-project/scouter on GitHub](https://github.com/scouter-project/scouter) --- Hysen Labs editorial analysis, written from the project's own repository and release notes. Cite the canonical page: https://hysenlabs.com/projects/scouter-project-scouter