Skip to content

Fix managed onboarding and invite unhosted drives over WebRTC (#1420) #50

Fix managed onboarding and invite unhosted drives over WebRTC (#1420)

Fix managed onboarding and invite unhosted drives over WebRTC (#1420) #50

name: Notify atomic-saas staging
# Every push to develop tells ontola/atomic-saas to redeploy its staging
# (staging.atomicserver.eu) against exactly this commit. Its
# deploy_staging.yml listens for the `atomic-server-develop` event and reads
# `client_payload.sha`. Until 2026-09-03 the same thing took a pin-bump PR
# over there plus a manual dispatch per change.
#
# Fires on the push, not on the main pipeline passing: PRs are green before
# they merge, the deploy verifies itself (health check, smoke test, rollback
# on the box), and waiting on the pipeline would tie staging to Mancave being
# awake. deploy_staging.yml in this repo (staging.atomicdata.dev) still waits
# for the pipeline; that is a different box with a different bar.
#
# ATOMIC_SAAS_DISPATCH_TOKEN: a fine-grained PAT for ontola/atomic-saas with
# Contents: read and write (repository_dispatch needs write). GITHUB_TOKEN
# cannot cross repositories. Missing token → the step fails loudly rather
# than staging silently going stale.
on:
push:
branches: [develop]
jobs:
dispatch:
runs-on: ubuntu-latest
timeout-minutes: 5
steps:
- name: Dispatch a staging deploy in atomic-saas
env:
GH_TOKEN: ${{ secrets.ATOMIC_SAAS_DISPATCH_TOKEN }}
run: |
set -euo pipefail
if [ -z "$GH_TOKEN" ]; then
echo "::error::ATOMIC_SAAS_DISPATCH_TOKEN is not set; atomic-saas staging will not follow this push."
exit 1
fi
gh api repos/ontola/atomic-saas/dispatches \
-f event_type=atomic-server-develop \
-f "client_payload[sha]=${{ github.sha }}" \
-f "client_payload[ref]=${{ github.ref }}"
echo "Dispatched: atomic-saas staging will build against ${{ github.sha }}."