fix(renovate): point the onboarding preset at the bitborg org #403

Sammanfogat

supernaut sammanfogade 2 incheckningar från fix/renovate-onboarding-preset-org in i main

2026-08-09 19:57:49 +00:00

supernaut kommenterad

2026-08-09 19:35:47 +00:00

Ägare

Two changes to the renovate role and this repo's Renovate config.

1. Stale org in the onboarding preset (the actual fix)

roles/renovate/templates/config.js.j2 still had:

onboardingConfig: { extends: ['local>gitborg/gitborg-docs'] }

This is a duplicated literal, not a templated value — nothing derives it from bitborg_domain — so the ADR 0039 §4b hostname move left it behind, pointing at a preset path that no longer exists.

The failure mode is silent and deferred, which is why it survived: only a newly onboarded repo reads onboardingConfig, and it would have started off-policy rather than erroring. Every existing repo has a committed renovate.json and never consults it. So there is no drift to correct on the eight current repos — this only matters for the next one added, which is exactly when nobody would think to check.

Needs an Ansible apply to reach the host; inert until then, and inert for existing repos either way.

While in there: renovate_git_author is still gitborg-renovate@gitborg.se. Left untouched, since its neighbours in that defaults file are explicitly marked LEGACY-PIN and it may well be deliberate — but if it is not, it is the same class of stale literal and worth a follow-up.

2. Why this repo does not automerge

Companion to bitborg-docs #81, which adds an opt-in automerge-safe preset. This repo deliberately does not extend it, and the reasoning is counterintuitive enough to be worth recording: it is excluded because merging here triggers nothing.

The customManagers bump *_image_tag values in group_vars — declared production state that reaches prod only on the next Ansible apply, which may well be run for an unrelated reason by someone who never reviewed the bump. Compounded by #63, where a tag bump can no-op without a pull+restart, so the running version does not reveal what happened either.

This repo's own devDependencies (prettier, markdownlint) would be safe to automerge, but are not worth a rule that risks being widened to image tags later.

The volume reduction from #81 — weekly batching, toolchain grouping across the npm and docker datasources — still reaches this repo through the shared preset.

Two changes to the `renovate` role and this repo's Renovate config. ## 1. Stale org in the onboarding preset (the actual fix) `roles/renovate/templates/config.js.j2` still had: ```js onboardingConfig: { extends: ['local>gitborg/gitborg-docs'] } ``` This is a **duplicated literal**, not a templated value — nothing derives it from `bitborg_domain` — so the ADR 0039 §4b hostname move left it behind, pointing at a preset path that no longer exists. The failure mode is silent and deferred, which is why it survived: only a *newly onboarded* repo reads `onboardingConfig`, and it would have started off-policy rather than erroring. Every existing repo has a committed `renovate.json` and never consults it. So there is no drift to correct on the eight current repos — this only matters for the next one added, which is exactly when nobody would think to check. Needs an Ansible apply to reach the host; inert until then, and inert for existing repos either way. While in there: `renovate_git_author` is still `gitborg-renovate@gitborg.se`. Left untouched, since its neighbours in that defaults file are explicitly marked `LEGACY-PIN` and it may well be deliberate — but if it is not, it is the same class of stale literal and worth a follow-up. ## 2. Why this repo does not automerge Companion to bitborg-docs #81, which adds an opt-in `automerge-safe` preset. This repo deliberately does not extend it, and the reasoning is counterintuitive enough to be worth recording: it is excluded *because* merging here triggers nothing. The `customManagers` bump `*_image_tag` values in `group_vars` — declared production state that reaches prod only on the next Ansible apply, which may well be run for an unrelated reason by someone who never reviewed the bump. Compounded by #63, where a tag bump can no-op without a pull+restart, so the running version does not reveal what happened either. This repo's own devDependencies (prettier, markdownlint) would be safe to automerge, but are not worth a rule that risks being widened to image tags later. The volume reduction from #81 — weekly batching, toolchain grouping across the npm and docker datasources — still reaches this repo through the shared preset.

supernaut lade till 2 incheckningar

2026-08-09 19:35:47 +00:00

fix(renovate): point the onboarding preset at the bitborg org 081cb44577

docs(renovate): record why infra image tags do not automerge

ci / ci (pull_request) Successful in 1m46s

Detaljer

52285df51e