Lesson acknowledged: Good Game

In the midst of our beta stage, we realized we didn’t really need a multi-AZ EKS cluster in a beta stage. This post is AWS-specific Long story short there is cross-AZ traffic, which gets very expensive very quickly when you have many databases/services talking with each other. Which was our case, of course. There are better solutions of course, like instructing k8s to deploy all pods talking with each other in the same AZ (and reducing the possible damage in case an entire AZ goes down). …Read More

For older cluster (I did not check yet with new ones), AWS used to default to GP2 for the Persistent Volumes. In most of the cases in my opinion, it makes sense to migrate to GP3 volumes, as those are cheaper and with higher performance in most of my usecases. This might not apply to everybody, so please check in the AWS documentation if that’s the case for you too! …Read More

The issue I’m a huge fan of admonitions (like this one below), but couldn’t find a proper way to implement them on top of a hugo theme (e.g. the one I’m using now) without touching it. Admonition Hi! I’m an admonition in Hugo in a theme that does not implement them (nor allows the import of CSS)! Shortcodes! My approach is based on Hugo shortcodes. Unfortunately, searching for info the only blog page I found was broken (and there was no working archived version), so here is what I did: …Read More

Everybody loves fancy graphs As an observability geek, I like collecting data to display it on fancy, useless graphs in my dashboard (possibly positioned below the useful ones). I tend to use Prometheus and Grafana Loki in my setups, to collect respectively metrics and logs. I like those two specifically because those are lightweight and powerful solutions. Unfortunately what’s trivial with logging/observability solutions like ELK (Elasticsearch, Logstash and Kibana, a chunky stack), can result in quite a bit of complexity in other platforms, like Prometheus and [Grafana] Loki. …Read More