Todo Microservice

There was a list of technologies I wanted to use for a project, and this todo app is a test to see how that would work.

This is just the server, but you can interact with it with GraphiQL, which is also served on localhost:8080/. As I'm going to use Elm for the frontend in said project, there will soon(-ish) be a client available in another repo for speaking with this server.

Contributing

If you want to, in any way, contribute to this experiment, or if you have questions about anything, feel free to open issues or PR's.

Technology/library roadmap

• go-kit
• graphql with graphql-go and granate
• paging of todos
• logging (go-kit)
• instrumenting (prometheus)
• opentracing (with appdash)
• client authentication
• authorisation (in progress)
• Proxying?
• Load balancing?
• Circuit breaking?
• Throtling?

Roadmap otherwise

• Todos
• Users
• Login

While I'm at it I want to try out a whole bunch of things. The todo service is exposed with graphql, but I want to expose it via a REST-ful api as well. This is both to see if the architecture would allow it without making compromises, and to test whether abstractions are done properly, or if logic leaks where it shouldn't. The REST stuff has low priority, though.

Running

$ cd $PROJECT_ROOT
$ go get . # Install dependencies
$ npm install # or `yarn`
$ gulp

Visit localhost:8080/ to interact with the server through GraphiQL. You can visit localhost:8080/metrics to see data from instrumentation. Use localhost:8080/graphql if you need to speak directly with the GraphQL API.

Inspiration

• go-kit/kit (shipping examle in particular)
• narqo/test-graphql

Resources

• graphql.org
• opentracing.io
• graphql-go
• granate
• Microservice pattern
• Hexagonal architecture