FOSDEM notes 2019
FOSDEM 2019 notes
This was my very first time at FOSDEM, although I’ve been following it since a while ago (thanks FOSDEM fellas for streaming all the main talks!). This year I had the chance to come over, thanks to my current employeer System73, along with some colleages. Main target this year for me was Infrastructure, Monitoring and Testing talks, and I had plenty of them (such as live talks, offline chats or even just following links from some slides). Here are my top notch talks:
Terraform
Two main talks about Terraform, both driven by Anton Babenko (such a good fella, main contributor on Terraform AWS modules): one about good practices on Terraform, and another as round table session about testing in terraform.
Good talk from Anton, where he exposed some good practices about reusing code, best practices and some tips writing Terraform code. He also showed us some interesting tools to incorporate into our ecosystem at System73 (see Tools section below). Best quote of the talk was this one:
- Testing in terraform
This one was more like an interactive chat (like open sessions in DevOps days) where Anton drove the session, but anyone could express and ask. It was a bit disappointed because I thought more testing stuff will be shown (in fact half of the room was against testing Terraform code, best way to test it is to actually apply it and see what happens), but on the other hand some good tips where given, like how to structure your modules to make them easy to reuse, reduce blast radius applying small changes and how to handle non-backward compatible changes from providers’ APIs.
Tools
-
Atlantis: it was introduced as a collaborative tool for developing and applying terraform code. It seems it allows teams to run terraform based on webhooks from pull requests, adding plan and apply results to code repository, thus increasing visibility and enabling collaboration within your team mates in an easy and standard way. Honestly, an interesting one, if we manage to use it well it will solve some of the problems we already have with terraform (code sharing, review, testing, etc).
-
Terragrunt: another interesting tool; this one helps keeping clean our terraform configurations, allowing us to work with terraform modules in a less messy way. I won’t be convinced until I give it a go, but pretty sure it will be a helping one given the fact we’re starting to have reusable code everywhere.
K8S
Almost everything on FOSDEM turned around Kubernetes, so many talks that I couldn’t cope with all of them. These are the ones I could attend and what I learned.
How we can craft custom K8S controllers with Ansible Operators to deploy and manage custom workload in a cluster. It seemed to me a bit tricky at the beginning, but after a while I got some pretty neat advantages automating your k8s controller with Ansible (easier to write, manage and customize resources if you already know Ansible).
I could only watch last 10 minutes, so not too much feedback from my side. Perhaps that made me wonder even more how Thanos works and if it makes sense to our organization (given the fact we have multiple Prometheus clusters world-wide), let’s see in a few weeks :)
Prometheus
Like Kubernetes, there was a huge hype about Prometheus, with some talks related to it. Again, I managed to listen to just one, but enough to convince me Prometheus is the future for metrics.
It was for me an overview of how to retrieve K8S metrics automatically and query them on Prometheus, with some tips and tricks around labelling and filtering. Very interesting if you’re new on Prometheus scrapping K8S metrics out-of-the-box. It also linked me to kube-prometheus repo, with all required stuff to deploy an end-to-end kubernetes cluster monitoring.
Extra stuff
Cool talk from our inspiring Kohsuke Kawaguchi about how to build a business around an open software project: Jenkins. He talked about all the different profitable steps Cloud Bees has followed until it became what it is nowadays. Besides, I managed to grab a brand new Jenkins X T-shirt!
I guess the main takeaway from this presentation was we are not far from what other people do with docs at System73: treat them as you will treat code. Applying same Agile principles (verify, build, deploy) in an automated fashion way not only QA or managers but devs will join keeping doc up to date without too much hassle.
- Brussels
FOSDEM is hosted in Brussels, one of the best european capitals I’ve ever visited, and it always welcomes FOSDEM with its traditional belgian fries, waffels, beers and this year, snow! Here you have some pictures and a video of our Brussels’ adventure.
What else?
After two intense days, more coming at Config Management Camp this week! For those who will attend it, see you there!