Networks, and servers are unreliable: they can fail, saturate… our requests should handle these scenarios. I created a small package in python — retry-requests —to do it for us easily.
The package requests and its underlying urrlib3 can already handle it, however it is not by default nor straightforward to configure.
The package retry-requests just writes the boilerplate so you don’t have to — and of course you can personalize it.