UPDATE Fri 16 Feb 2018 : I’m back from Serverlessconf Paris 2018 where I learned about observability of a distributed system with the help of a fantastic tool : IOPipe. I suggest you take some time to give it a try, it is awesome.
Last year I wrote a blog post about the urge to use a distributed logging system to pinpoint outages in cloud applications.
This is very frustrating to me because it seems we’ve no easy way of finding what’s wrong, yet. But hey that’s life, we’re improving.
This is the perfect time for me to remind you about best practices. So I have to say this:
- Write logs to the standard output in your app/framework/API/proxy about what you’re doing, catch errors, trace them and rethrow. You can’t rely on your consumer to do that on your behalf.
- Comment your open source code. The community is here to help. Doing so gives us the ability to easily understand your code.
Don’t break the chain, give us a bridge to safely cross the water.
Happy coding 🙂