@mgfx/inspector v0.1.1

@mgfx/inspector

A tool for exploring, analysing and debugging mgFx runtime activity across time and space.

Features

• Explore mgFx Task executions across time on the horizontal axis.
• Tasks are grouped vertically by their mgFx context.
• For a given task: examine the state, duration, input arguments and output result.
• View activity timing for Contexts (tasks in current Context vs. tasks in descendant Contexts).
• View application activity in real-time (experimental).

Usage

1. Configure your mgFx application to use the instrumentation API.

   For example:

   import { Instrumenter } from 'mgfx';
   
   new Instrumenter({ scheduler: yourMgFxSchedulerInstance });

2. Run your mgFx application, passing the instrumentation output to the mgFx recorder.

   For example:

   $ node your-mgfx-app.js | npx @mgfx/recorder sqlite /path/to/recorder/output.db

3. While your application is running, start the mgFx Inspector, specifying the path to the same sqlite database that the mgFx recorder is using:

   $ npx @mgfx/inspector sqlite /path/to/recorder/output.db

4. The Inspector user interface may be accessed via http://localhost:8080. The port that the HTTP server starts on may be changed via the --port command-line option.

Experimental Features

Realtime activity ('follow' mode)

This feature has not yet been surfaced in the UI, but may be enabled by opening the browser window's developer tools and running:

window.FOLLOW()

in the JavaScript console.

Roadmap/Planned Features

This list is by no means exhaustive; please feel free to raise an issue suggesting a feature or a Pull Request implementing one.

• Stress-test: Ensure that the Inspector performs adequately on large data sets.
• Colour-code Tasks in the Timeline based on their state.
• Navigation controls ('go to most recent', 'go to last rejected Task', etc).
• Support for PostgreSQL as an additional database provider.
• Production Usage: Document how to use the Recorder and Inspector for analysis and monitoring of an application running in Production.
• 'Real' realtime mode using WebSockets.
• View the 'chain' of execution for a given Task. For example, highlight ancestors, descendants and siblings of a Task when hovering over it.
• Better display/interaction of short-lived tasks without the need to zoom to down to milliseconds.