A common practice with Docker containers is to log everything they output to /dev/stdout and /dev/stderr. This is perfectly fine for most containers. But with dockerized load-balancers, I can create issues. This is what we recently faced at Hunter.

Over a year ago (I can’t believe how fast time flies!), I began working 100% remotely for hunter.io. This has been my best professionnal decision ever. In this post, I wanted to share a couple of things working remotely has taught me.

At Hunter, we recently decided to move parts of our containers management from Cloud66 to Kubernetes. We wanted to have more control over the resources requested by our apps, and also speed up the deployment process which was really slow at C66.

These are the slides of a talk I gave for the december 2016 Grenoble Human Talks.

Docker became in a couple of years the DevOps’ new best friend. The goal of this article is not to present the multiple benefits of Docker, but rather to talk about an use case where Docker can really be useful: Zero-downtime deployments. In other words, the ability to update your online dockerized service (an app, a website, etc.) without any downtime.