Well, that went better than expected :)
I’ve given my whole public-facing infrastructure a belated spring-clean. I have now largely moved off my collection of VPS systems provisioned using Ansible, to a stack much more in-line with what I help customers build at my job. I’m now hosted on a managed Kubernetes cluster provided by DigitalOcean, and everything from this blog to the infrastructure automation is expressed as code in git repositories and rolled out using Concourse.
The whole thing has all been torn down and recreated from scratch several times a day with no issues so I’m feeling pretty pleased with it. I also have:
- SSL Ingress through Traefik
- Monitoring through Prometheus & Grafana
- ELK stack for logs
Publishing this post is now a simple matter of committing the updates to git and then waiting a few minutes for my Concourse pipelines to pick up the change, build new containers and deploy. All of which means I now have far less excuses not to write more!