Hacker News new | past | comments | ask | show | jobs | submit login

The author is doing it wrong - they don’t need to run a local k8s cluster with 150 services - this is a monolith way and they should have stayed with mononlyth if they want to do this.

Microservices require quite a bit of dev setup to get it right but often it comes down to be able to run a service locally against a dev environment, that has all those 150 other microservicea already running.

Queues are setup to be able to route them to your local workstation, local ui should have ability to proxy to ui running in dev (so that you don’t run entire amazon.com or such locally), deployments to dev have to be all automated and largely lights out, and so on.... it takes a bit of time to get these dev things right, but it doesn’t require running entire environment locally just to write a few lines of code.

Debugging and logging/tracing are an issue - but these days there are some pretty good solutions to that too - Splunk works quite well, and saves a lot of time tracking issues down.




For tracing, I tried Jaeger recently and it looks promising! https://www.jaegertracing.io/




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: