You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
@schutera — surfaced while documenting the OTA release process (#151). The firmware-OTA side is now covered; this is the one unresolved operational question it exposed, and it's yours to decide.
Situation
The deployment docs say the web services deploy from the production branch (docs/07-deployment-view/production-deployment.md, production-runbook.md), but that branch no longer matches what's live:
origin/production is 151 commits behind main and 25 ahead (divergent history), last touched 2026-05-02 (bf8b314).
So the documented services deploy source ≠ reality.
Why it matters
The services deploy source is ambiguous, and the new OTA release runbook's publish step (rebuild the frontend image) rides this same track — so the ambiguity leaks into firmware releases too. Firmware-OTA versioning itself is unaffected (it's cut on main + prod-* tags).
Options — pick one
Adopt main as the services deploy source (matches reality). Update the two deploy docs to clone/pull main; archive or delete production. → First confirm the 25 commits unique to production are obsolete / already squashed into main and nothing is genuinely unmerged.
Keep production as the deploy branch. Fast-forward it onto main now and make "update production" a step in every services deploy. → Same caveat: review the 25 divergent commits first.
Which model is actually in use? Once you pick, I can do the doc updates and/or the branch cleanup.
Until then, the runbook and the chapter-11 entry both carry a "verify your actual deploy source before relying on this" hedge — this issue is what gives that hedge a finite lifespan.
Refs: docs/07-deployment-view/firmware-release.md → "Git: branch & tag model"; docs/11-risks-and-technical-debt/README.md → "production branch drifted from the deployed services"; #151.
@schutera — surfaced while documenting the OTA release process (#151). The firmware-OTA side is now covered; this is the one unresolved operational question it exposed, and it's yours to decide.
Situation
The deployment docs say the web services deploy from the
productionbranch (docs/07-deployment-view/production-deployment.md,production-runbook.md), but that branch no longer matches what's live:origin/productionis 151 commits behindmainand 25 ahead (divergent history), last touched 2026-05-02 (bf8b314).main-only code: the security: production API key is baked into the public JS bundle; admin login gate is health-only #142 admin session (backend/src/session.ts/POST /api/admin/login) exists onmain, is absent onproduction, and responds in production.So the documented services deploy source ≠ reality.
Why it matters
The services deploy source is ambiguous, and the new OTA release runbook's publish step (rebuild the frontend image) rides this same track — so the ambiguity leaks into firmware releases too. Firmware-OTA versioning itself is unaffected (it's cut on
main+prod-*tags).Options — pick one
mainas the services deploy source (matches reality). Update the two deploy docs to clone/pullmain; archive or deleteproduction. → First confirm the 25 commits unique toproductionare obsolete / already squashed intomainand nothing is genuinely unmerged.productionas the deploy branch. Fast-forward it ontomainnow and make "updateproduction" a step in every services deploy. → Same caveat: review the 25 divergent commits first.productionis an intentional staging / release-gated branch. Then document the promotion gate and why it lags — though the current evidence (prod runningmain-only security: production API key is baked into the public JS bundle; admin login gate is health-only #142) argues against this.Ask
Which model is actually in use? Once you pick, I can do the doc updates and/or the branch cleanup.
Until then, the runbook and the chapter-11 entry both carry a "verify your actual deploy source before relying on this" hedge — this issue is what gives that hedge a finite lifespan.
Refs:
docs/07-deployment-view/firmware-release.md→ "Git: branch & tag model";docs/11-risks-and-technical-debt/README.md→ "productionbranch drifted from the deployed services"; #151.