I setup Gitea + Drone on 2GB Vultr VPS. Is very good for small project. Then I migrate work monorepo. 3GB node_modules. CI build start, server die. OOM every time. I think "maybe Woodpecker more light" and almost switch.
But actually is my fault. The repo is monster. I check with `du -sh` and node_modules is 3.1GB. Build need 4GB+ ram. No CI tool fix this. I split to git submodules, cache node_modules in Hetzner S3, now build pass.
Lesson: check your repo before blame the tool. I make this test too late.
The woodpecker is lighter but ur right its not magic. I run gitea on 512MB at RackNerd for $8/year. No ci tho. Definitely separate runner if u need build. Im too cheap for that lol
Do you actually run Gitea on 1GB or just recommend it
I run it. My personal Gitea sits on a Contabo VPS S (4GB) in Nuremberg, but the instance itself idles around 400MB. I have seen stable 1GB deployments in our Frankfurt datacenter for small teams (under 10 repos, light traffic). The 1GB recommendation assumes you are NOT also running CI on the same node.
If you try Gitea + anything else on 1GB you will have a bad time. The kernel OOM killer does not negotiate.
The 1GB recommendation assumes you are NOT also running CI on the same node
This. I learned this the hard way with my 512MB RackNerd box. Gitea alone is fine. Add one Drone build and suddenly I am SSHing in at 3am to reboot. Never again.
Separate runner is the way. I run Gitea on Vultr 1GB in London, Drone runner on Hetzner CPX11 in Falkenstein. Latency between them is fine for git fetch. Total cost under 10 EUR, builds never kill my repo.
Post a reply
You need an account to reply.
Log in or
register to join the conversation.