← all posts site ↗

retiring the docker stacks i never touch

it started with the pdf service. it's a fat consumer of ram and cpu for a tool i barely touch, and asking why that one was so heavy opened the floodgate: i have a lot of shit on this system i don't actually use. the pdf job got swapped for a tool that does the same work in a fraction of the footprint, and the uptime monitor that kept sending down-and-up notifications for everything got replaced with a tiny health checker that watches the endpoints and says nothing unless something matters, twenty something megs all in.

then the purge snowballed. the download web ui went. the update detector went. the fitness app mirror went. the travel planner went. the link unblocker went, along with the cloudflare bypass helper that lived inside it, after checking that nothing else on the box referenced it anywhere. the reverse proxy hosts for the dead services got deleted, the health checker endpoints got pruned down to the ones that exist, and the dashboard widgets for the ghosts got removed, which is why they had still been visible after the containers died.

the verification part was the satisfying part. every remaining service answered after the proxy cleanup, the health checker came up green on all twenty three endpoints it watches, and the whole day of work stuck to one principle: same box, same job, but a couple dozen megabytes of containers instead of gigabytes, plus a retired folder of backups waiting for the final purge. the best thing about killing services is noticing how few of them you could name from memory before you started.