rizalgowandy / statsviz

Instant live visualization of your Go application runtime statistics (GC, MemStats, etc.) with a single import, à la `net/http/pprof` :rocket:

Statsviz

Instant live visualization of your Go application runtime statistics (GC, MemStats, etc.).

• Import "github.com/arl/statsviz"
• Register statsviz HTTP handlers
• Start your program
• Open your browser at http://host:port/debug/statsviz
• Enjoy...

How does that work?

Statsviz serves 2 HTTP handlers.

The first one (by default /debug/statsviz) serves an html/js user interface showing some initially empty plots.

When you points your browser to statsviz user interface page, it connects to statsviz second HTTP handler. This second handler then upgrades the connection to the websocket protocol and starts a goroutine that periodically calls runtime.ReadMemStats, sending the result to the user interface, which inturn, updates the plots.

Stats are stored in-browser inside a circular buffer which keep tracks of a predefined number of datapoints, 60, so one minute-worth of data, by default. You can change the frequency at which stats are sent by passing SendFrequency to Register.

Usage

go get github.com/arl/statsviz

Either Register statsviz HTTP handlers with the http.ServeMux you're using (preferred method):

mux := http.NewServeMux()
statsviz.Register(mux)

Or register them with the http.DefaultServeMux:

statsviz.RegisterDefault()

If your application is not already running an HTTP server, you need to start one. Add "net/http" and "log" to your imports and the following code to your main function:

go func() {
    log.Println(http.ListenAndServe("localhost:6060", nil))
}()

By default the handled path is /debug/statsviz/.

Then open your browser at http://localhost:6060/debug/statsviz/

Plots

On the plots where it matters, garbage collections are shown as vertical lines.

Heap

MSpans / MCaches

Size classes heatmap

Objects

Goroutines

GC/CPU fraction

Examples

Have a look at the _example directory to see some different ways to use Statsviz, such as:

• using http.DefaultServeMux
• using your own http.ServeMux
• wrap HTTP handler behind a middleware
• register at /foo/bar instead of /debug/statviz
• use https:// rather than http://
• using with various Go HTTP libraries/frameworks:
  • fasthttp
  • gin
  • and many others thanks to wonderful contributors!

Contributing

Pull-requests are welcome! More details in CONTRIBUTING.md

Roadmap

• add stop-the-world duration heatmap
• increase data retention
• light/dark mode selector
• plot image export as png
• save timeseries to disk
• load from disk previously saved timeseries

Changelog

See CHANGELOG.md.

License

• MIT License