Paul CI worker available: one additional concurrent job #41

New issue

Open

opened 2026-10-29 15:13:52 +00:00 by pedro · 0 comments

pedro commented

2026-10-29 15:13:52 +00:00

Owner

A shared Forgejo CI worker named paul-ci is now available on physical Paul (10.10.1.12). It adds one concurrent job alongside the existing PVE1 runners.

Existing divesim, track-pt3r (AdminiSync), and ansible-control (the-lair) jobs can automatically run on either matching worker, with their existing job images.
Other Pedro-owned projects can opt in with runs-on: linux-x64-ci (Ubuntu 24.04 Actions image). Use runs-on: paul-ci only when explicitly targeting Paul.
Each job is limited to 2 CPUs / 3 GiB RAM, with a 45-minute timeout. The Docker daemon and service containers share a 2-CPU/4-GiB budget; runner overhead adds 0.25 CPU/256 MiB. Only one job runs on this worker at a time.
Scratch is local and disposable (16 GiB Docker data); no persistent cache, GPU, host Docker socket, deployment credentials, or Kubernetes API token is provided. Normal repository Actions secrets still follow Forgejo's per-job behavior.
This owner-scoped worker is for trusted Pedro-owned projects. Keep public-fork approval gates and deployment locks; splitting independent checks into separate jobs can use both machines, while sequential steps in one job cannot.

Provisioning, labels, maintenance and trust boundaries: runner runbook.

This is a resource-availability notice; no project workflow or deployment behavior was changed by this notification.

Passed worker smoke test: checkout, PostgreSQL networking, effective job limits, and credential-mount checks.

A shared Forgejo CI worker named `paul-ci` is now available on physical Paul (10.10.1.12). It adds **one concurrent job** alongside the existing PVE1 runners. - Existing `divesim`, `track-pt3r` (AdminiSync), and `ansible-control` (the-lair) jobs can automatically run on either matching worker, with their existing job images. - Other Pedro-owned projects can opt in with `runs-on: linux-x64-ci` (Ubuntu 24.04 Actions image). Use `runs-on: paul-ci` only when explicitly targeting Paul. - Each job is limited to **2 CPUs / 3 GiB RAM**, with a **45-minute timeout**. The Docker daemon and service containers share a 2-CPU/4-GiB budget; runner overhead adds 0.25 CPU/256 MiB. Only one job runs on this worker at a time. - Scratch is local and disposable (16 GiB Docker data); no persistent cache, GPU, host Docker socket, deployment credentials, or Kubernetes API token is provided. Normal repository Actions secrets still follow Forgejo's per-job behavior. - This owner-scoped worker is for trusted Pedro-owned projects. Keep public-fork approval gates and deployment locks; splitting independent checks into separate jobs can use both machines, while sequential steps in one job cannot. Provisioning, labels, maintenance and trust boundaries: [runner runbook](https://git.guresky.com/pedro/the-lair/src/branch/main/docs/FORGEJO_RUNNERS.md). This is a resource-availability notice; no project workflow or deployment behavior was changed by this notification. [Passed worker smoke test](https://git.guresky.com/pedro/the-lair/actions/runs/384): checkout, PostgreSQL networking, effective job limits, and credential-mount checks.