mjpitz/go-gracefully

A simple, efficient, and easy to use Golang library for implementing application health checks. Features first-class support for stream-based checks.

👋 Welcome! My name is Mya.

I'm an Engineering Director who enjoys building and working with teams around the world. Outside of work, I'm an avid hockey player and enjoy spending time in my garden. I'm passionate about doing more to save our environment ♻️. While much of my learning has been confined to my own space, I look forward to sharing my knowledge with others.

📖 Favorite Books 📖

Hold Me Tight - Dr. Sue Johnson EdD - [Amazon] [Kindle] [Audible]
Superbosses - Sydney Finkelstein - [Amazon] [Kindle] [Audible]
Dare to Lead - Brené Brown - [Amazon] [Kindle] [Audible]
Team Topologies - Matthew Skelton and Manuel Pais - [Amazon] [Kindle] [Audible]

📰 Recent Blog Posts 📰

Bye-Bye Twitter - January 17th 2022
Extending my Security System with a Raspberry Pi Network Bridge - June 16th 2021

⚒️ Tools and Technologies ⚒️

                       

All projects adopt the Contributor Covenant Code of Conduct.

go-gracefully's Issues

Examples

Let's add some examples to show people how to get started. The two that come to mind immediately are:

• how to hook into gRPC
• how to hook into HTTP mux

I really liked the way grpc-go set this up and would like to follow the example they set. The directory structure is:

examples/
  |- features/
    |- periodic-check
    |- stream-check
    |- comparative-check
    |- error-rate-check
  |- integrations/
    |- grpc
    |- http

Each directory contains a main.go file demonstrating the given feature / integration and a README.md that describes how to run it (mostly using go run.

Allow failure thresholds to be configurable

Right now, I believe this is static in code. The idea is to let consumers configured when the server returns an error vs ok.

Add error-rate check

The idea would be to add a new type of check that allows errors to be pushed into a channel. The associated check keeps track of the error rates and compares it with the desired thresholds. When it reaches / passes a given threshold, it changes the checks state.

Deprecate in favor of indeedeng/libhealth

I wrote this as a placeholder while I was waiting for the aforementioned project to be open sourced. It would be nice if the JSON data returned wasn't so verbose, but that's an easy PR.

Reduce go routine usage

The current version was written in a way to get a solution out the door pretty quickly. It would be good to revisit and reduce the number of goroutines that are needed here.

Reduce goroutine requirements

For simplicity in early development, I gave each check their own goroutine to feed the shared channel. Since goroutines are much lighter weight than Java threads, I didn't (and honestly still don't) think too much about this.

Curious if this is something folks will actually need for their system. If so, it would be good to get some input from folks.

Raise test coverage

Now that I've gotten codecov for the project, I want to raise the minimum test coverage.

Some errors not reporting correctly..

This was kinda odd to see today. When one of my gRPC services returned an error, the message didn't actually render in the health report.

"last_known_check": {
  "state": "outage",
  "error": {},
  "timestamp": "2020-12-10T16:37:23.1725756Z"
}