Forum

Hi David,

We took a look at your RustyServer (Hetzner cpx11 — 2 vCPU / 2 GB). It's been crash-looping for the last day, and the root cause is straightforward: the server is severely undersized for what's running on it.

Live snapshot from when we checked:

• Load average: 45–72 (on a 2-core box — should be ≤ 2)
• Free memory: ~84 MiB out of 1.9 GiB total, no swap
• Every Zeabur system pod (cadvisor, fluent-bit, ingress, nats, dns, kube-watch, etc.) is constantly restarting due to OOM / IO pressure
• We see the Claude Code CLI process (claude) and a separate bot process (whale_b...) running on the host alongside k3s — together they consume a meaningful chunk of the already-tiny RAM budget

You don't currently have any Zeabur services deployed on this server, so all of the load is k3s + the things you're running on the host directly. K3s alone needs about 500 MB of RAM headroom to be stable, and once you add Zeabur's system components and your host processes, 2 GB simply runs out.

Recommendation: resize the server to at least cpx21 (3 vCPU / 4 GB) or larger. You can do this from Servers → RustyServer → Resize. After the resize, the K3s "offline" status should clear automatically once the system pods stop crash-looping.

If you also want to keep running the Claude Code CLI and the bot on the host alongside Zeabur, you'll likely want even more headroom — cpx31 (4 vCPU / 8 GB) is a safer fit for that workload.

Let us know after the resize and we'll re-check to make sure everything is healthy.

I upgraded my server I have 2 questions:

1. Can you confirm that I created my Service correctly for this server? Something seems off like I didn't do it correctly.
2. In what instances, would I be charged overages by using your server? I was doing research and it stated that Zeabur can charge overages based on usage.

Thank you!

David

Will you be able to give me a discount or credit on my account? I'm still waiting on a response and its been well over 12 hours?

Is anyone online?

Hi David, sorry for the delayed response.

Good news — your server upgrade worked. It's now running 4 vCPU / 7.6 GB RAM with load near zero and all system pods healthy. The K3s offline issue is fully resolved.

1. Your service setup

Your rusty-service currently has no running pods. It's configured as a LOCAL source with the command cd ~/whale_bot && cargo run --release — this won't work because the Zeabur container doesn't have access to your host's home directory or a Rust toolchain.

To deploy your Rust bot on Zeabur, the recommended approach is:

• Push your code to a GitHub repo and connect it as a Git source — Zeabur will auto-detect and build Rust projects
• Or create a Dockerfile and deploy as a Docker image

If you'd like help setting up either approach, let us know.

2. Overages

Since your project is on a dedicated server, you're only charged the flat monthly server fee. There are no per-usage memory/egress/storage overages — those only apply to services on Zeabur's shared infrastructure.

Let us know how you'd like to proceed with the service setup.

I'm still testing so should I wait to deploy Rustbot to Zeabur? And do you have any links to documentation where I can read about this?

Hi David,

Yes — if you're still testing your bot locally, it makes sense to keep developing until you're happy with it, then deploy to Zeabur when it's ready.

When you're ready to deploy, here's the documentation on how to deploy projects on Zeabur:

• Deploying from GitHub: https://zeabur.com/docs/deploy/github
• Deploying with Docker image: https://zeabur.com/docs/deploy/docker

Zeabur auto-detects Rust projects from GitHub repos, so the easiest path is to push your code to GitHub and connect the repo.

Let us know if you need anything else!

This post has been inactive for a while. We will be closing it in 2 days if there is no new activity.