Due to ahem.. "issues" related to a network change in the DC, that's going to balloon out to around 10 or 12 for a little while, then shrink back down.
- 2x load balancers (1 is active for public traffic; backup for private traffic; 1 is active for private traffic; backup for public traffic);
- 3x DB cluster nodes (currently at 5 due to aforementioned issues);
- 2x Web/App nodes
Currently planning to adapt this into 3 distinct sites with simpler 1x web/app + 1x db per-site to move the redundancy point out to specific DCs rather than to individual roles within a single DC.
Litestream backs up the databases automatically to S3 compatible storage, and linode offers automatic snapshots of the VPS.
My deployment process is git push and that’s it. If I need to setup a new webapp, I just clone the repo, add to supervisor and caddy config, and I’m good to go.
If I need to do CI, I use GitHub, which just git pushes to the server to deploy.
Everything running over CloudFlareTunnel.
It's boring technology - PHP+MySQL+little bit Node.JS
Are the applications running on it CPU intensive?
Power usage on idle is around 70-80 Watt, on normal usage is around 100W, on peak can go to 200-250W.
Apps are not CPU intensive except one that uses disk in inefficient way. That's why i make a ram drive (tmpfs), put data there and then get out.
Production is Argo+Helm+Kubernetes+public cloud provider
Development is bash+docker compose+local computer or remote docker host