Transfer cluster repository to Personal organization with a GitOps-safe migration runbook #76

Closed
opened 2026-11-07 22:16:15 +00:00 by martin · 4 comments
Owner

Goal

Transfer the existing Forgejo repository from martin/cluster to the confirmed Personal organization, resulting in Personal/cluster, without interrupting GitOps reconciliation or losing repository history and issue tracking. Document a repeatable migration runbook and rollback procedure before executing the transfer.

This issue schedules the work; creating it does not authorize performing the transfer or changing the live cluster.

Repository-specific context

  • modules/common.nix defines networking.clusterRepo = "https://code.${domain.root}/martin/cluster.git" and uses it for nixidy.target.repository; the target branch defaults to main.
  • env/dev/argocd.nix uses clusterRepo for Argo CD's repositories.cluster.url.
  • Shared Nixidy defaults enable automatic synchronization and pruning. An incorrect source or manifest path can affect all environments, including Argo CD's own deployment.
  • The local SSH remote currently points to ssh://forgejo@nidaros.secure.martials.no/martin/cluster.git; distinguish this SSH endpoint from the HTTPS GitOps URL.
  • networking.registry separately points to code.${domain.root}/martin. Repository transfer must not implicitly relocate container packages or break image pulls; assess package ownership separately.
  • Coordinate with #66 and #70 if infrastructure/Argo CD environment migrations are underway; avoid combining unrelated cutovers.

Required migration runbook

1. Inventory and preparation

  • Confirm transfer permissions and organization team access, and verify that Personal/cluster does not already exist. Preserve visibility, default branch, branch protections, collaborators, and bot access.
  • Inventory all references to the old repository in Nix source, generated manifests, bootstrap/root Applications, ApplicationSets, AppProjects' sourceRepos, repository registrations, URL-scoped credential templates, external scripts, docs, badges, and local clones.
  • Inspect the live Argo CD Application/ApplicationSet/AppProject configuration, not only Git. Record source URL(s), revision, manifest path(s), destinations, automation/pruning settings, and current revision/health for every environment.
  • Inventory Woodpecker registration, repository-local and global pipelines, cron jobs, webhook URLs/secrets, repository IDs/slugs, status checks, write-back credentials, and Renovate discovery/configuration. Confirm organization membership and token/deploy-key scopes still allow fetches and pushes after transfer.
  • Back up Git refs and repository metadata as appropriate; record a known-good commit and securely capture relevant live configuration and current automation settings. Do not expose secret values in the issue or commit them.
  • Prepare and review a PR changing the canonical GitOps URL to https://code.martials.no/Personal/cluster.git, plus any applicable allowlists and credential registrations. Generate all affected manifests through Nixidy; never hand-edit generated YAML or encrypted SOPS files.
  • Run just fmt, just test, just test-IT, and builds for dev, infra, and prod; review the generated diff for URL-only changes and unintended deletion/pruning, namespace, or application-identity changes. The user signs the commits.

2. Controlled cutover

  • Schedule a maintenance window, freeze unrelated merges, and pause CI/Renovate write-back and relevant Argo CD automated reconciliation/pruning. Record how to restore each setting, and ensure a parent Application/ApplicationSet cannot immediately undo the temporary pause. Wait for active syncs/pipelines to finish.
  • Use Forgejo's native repository transfer, not creation of a replacement repository, so history, issues (including the migrated Beads records), PRs, and other metadata remain attached to the same repository. Complete any transfer acceptance required by Forgejo.
  • Verify direct HTTPS fetch access to the new URL from Argo CD's execution context and SSH/HTTPS access for maintainers and bots. Do not rely on the old URL redirect, especially for SSH, allowlist checks, or URL-scoped credentials.
  • Restore/adjust Woodpecker registration, webhooks, Renovate access, team permissions, deploy keys, and branch-protection status checks as needed. Verify rather than assume these survive transfer automatically.
  • Merge/push the prepared canonical-URL and generated-manifest changes to the transferred repository. Update origin to the new SSH path. Publish the complete source and generated manifests before enabling reconciliation against them.
  • Use the smallest documented live source/credential/AppProject update necessary to recover bootstrap/root reconciliation when it cannot discover its own new URL. Ensure all such changes match committed desired state; temporary live edits are not the final source of truth.
  • Refresh and perform controlled synchronization of bootstrap/root applications and then dependent applications for each environment. Inspect the diff before applying; stop on unexpected deletions or missing manifests. Restore automatic sync/pruning and CI/Renovate only after successful validation.

3. Verification and documentation

  • Confirm Argo CD can fetch the canonical new URL and each application's expected revision/path; no repository/authentication/allowlist errors, unexpected pruning, or migration-caused degraded applications remain.
  • Confirm every environment reaches its expected Synced/Healthy state (document any pre-existing exceptions), Argo CD can still manage itself, and a small reviewed follow-up change reconciles from the transferred repository.
  • Trigger a safe CI verification run and verify webhook delivery, required status checks, authorized manifest write-back, and Renovate discovery at Personal/cluster.
  • Verify history/refs, open and closed issues, PRs, labels, protections, visibility, access, and package/image-pull behavior. Update AGENTS.md, contributor/bootstrap docs, and external references to the new owner.
  • Commit the final runbook with exact discovered resource names/commands, ordering, permissions, verification evidence, and recovery instructions. Remove dependence on old-URL redirects; clearly identify historical references that intentionally remain.

Rollback and stop conditions

  • Stop and keep automation paused on fetch/authentication failures, unexpected pruning, missing applications, or failed critical health checks. Existing workloads must not be deleted to work around a repository-access problem.
  • Document whether reverse transfer to martin is permitted and whether the old namespace remains available before cutover. If reversing transfer, restore working repository access first, then revert canonical URLs/allowlists/credentials and regenerate manifests from the known-good Git state.
  • If reverse transfer is unavailable, recover repository access at Personal/cluster and roll back configuration there; do not restore a source URL that no longer hosts an accessible repository.
  • Document secure recovery of bootstrap/root source settings and the exact restoration of paused automation. Temporary live recovery changes must be reconciled back into Git. Re-verify all environments before resuming automation.

Acceptance criteria

  • A reviewed runbook describes preparation, ordered cutover, verification, stop conditions, and rollback, including the self-managed Argo CD bootstrap dependency.
  • The existing repository is transferred to Personal/cluster with history, issues, PRs, permissions, and protections verified.
  • Nix source and generated manifests for all affected environments use the canonical new URL; no generated YAML or encrypted SOPS values were hand-edited.
  • Argo CD reconciles all environments from the new repository with no unexpected pruning or migration-caused downtime; original automation settings are restored.
  • Woodpecker, Renovate, webhooks, status checks, and required fetch/write credentials work under the new organization.
  • Container packages and image pulls remain functional; registry relocation is not silently bundled into this task.
  • Documentation, bootstrap instructions, and local-remotes guidance are updated, and verification evidence plus the tested/validated recovery plan are recorded.
## Goal Transfer the existing Forgejo repository from `martin/cluster` to the confirmed `Personal` organization, resulting in `Personal/cluster`, without interrupting GitOps reconciliation or losing repository history and issue tracking. Document a repeatable migration runbook and rollback procedure before executing the transfer. **This issue schedules the work; creating it does not authorize performing the transfer or changing the live cluster.** ## Repository-specific context - `modules/common.nix` defines `networking.clusterRepo = "https://code.${domain.root}/martin/cluster.git"` and uses it for `nixidy.target.repository`; the target branch defaults to `main`. - `env/dev/argocd.nix` uses `clusterRepo` for Argo CD's `repositories.cluster.url`. - Shared Nixidy defaults enable automatic synchronization **and pruning**. An incorrect source or manifest path can affect all environments, including Argo CD's own deployment. - The local SSH remote currently points to `ssh://forgejo@nidaros.secure.martials.no/martin/cluster.git`; distinguish this SSH endpoint from the HTTPS GitOps URL. - `networking.registry` separately points to `code.${domain.root}/martin`. Repository transfer must not implicitly relocate container packages or break image pulls; assess package ownership separately. - Coordinate with #66 and #70 if infrastructure/Argo CD environment migrations are underway; avoid combining unrelated cutovers. ## Required migration runbook ### 1. Inventory and preparation - Confirm transfer permissions and organization team access, and verify that `Personal/cluster` does not already exist. Preserve visibility, default branch, branch protections, collaborators, and bot access. - Inventory all references to the old repository in Nix source, generated manifests, bootstrap/root Applications, ApplicationSets, AppProjects' `sourceRepos`, repository registrations, URL-scoped credential templates, external scripts, docs, badges, and local clones. - Inspect the **live** Argo CD Application/ApplicationSet/AppProject configuration, not only Git. Record source URL(s), revision, manifest path(s), destinations, automation/pruning settings, and current revision/health for every environment. - Inventory Woodpecker registration, repository-local **and global** pipelines, cron jobs, webhook URLs/secrets, repository IDs/slugs, status checks, write-back credentials, and Renovate discovery/configuration. Confirm organization membership and token/deploy-key scopes still allow fetches and pushes after transfer. - Back up Git refs and repository metadata as appropriate; record a known-good commit and securely capture relevant live configuration and current automation settings. Do not expose secret values in the issue or commit them. - Prepare and review a PR changing the canonical GitOps URL to `https://code.martials.no/Personal/cluster.git`, plus any applicable allowlists and credential registrations. Generate all affected manifests through Nixidy; never hand-edit generated YAML or encrypted SOPS files. - Run `just fmt`, `just test`, `just test-IT`, and builds for `dev`, `infra`, and `prod`; review the generated diff for URL-only changes and unintended deletion/pruning, namespace, or application-identity changes. The user signs the commits. ### 2. Controlled cutover - Schedule a maintenance window, freeze unrelated merges, and pause CI/Renovate write-back and relevant Argo CD automated reconciliation/pruning. Record how to restore each setting, and ensure a parent Application/ApplicationSet cannot immediately undo the temporary pause. Wait for active syncs/pipelines to finish. - Use Forgejo's **native repository transfer**, not creation of a replacement repository, so history, issues (including the migrated Beads records), PRs, and other metadata remain attached to the same repository. Complete any transfer acceptance required by Forgejo. - Verify direct HTTPS fetch access to the new URL from Argo CD's execution context and SSH/HTTPS access for maintainers and bots. Do not rely on the old URL redirect, especially for SSH, allowlist checks, or URL-scoped credentials. - Restore/adjust Woodpecker registration, webhooks, Renovate access, team permissions, deploy keys, and branch-protection status checks as needed. Verify rather than assume these survive transfer automatically. - Merge/push the prepared canonical-URL and generated-manifest changes to the transferred repository. Update `origin` to the new SSH path. Publish the complete source and generated manifests before enabling reconciliation against them. - Use the smallest documented live source/credential/AppProject update necessary to recover bootstrap/root reconciliation when it cannot discover its own new URL. Ensure all such changes match committed desired state; temporary live edits are not the final source of truth. - Refresh and perform controlled synchronization of bootstrap/root applications and then dependent applications for each environment. Inspect the diff before applying; stop on unexpected deletions or missing manifests. Restore automatic sync/pruning and CI/Renovate only after successful validation. ### 3. Verification and documentation - Confirm Argo CD can fetch the canonical new URL and each application's expected revision/path; no repository/authentication/allowlist errors, unexpected pruning, or migration-caused degraded applications remain. - Confirm every environment reaches its expected Synced/Healthy state (document any pre-existing exceptions), Argo CD can still manage itself, and a small reviewed follow-up change reconciles from the transferred repository. - Trigger a safe CI verification run and verify webhook delivery, required status checks, authorized manifest write-back, and Renovate discovery at `Personal/cluster`. - Verify history/refs, open and closed issues, PRs, labels, protections, visibility, access, and package/image-pull behavior. Update `AGENTS.md`, contributor/bootstrap docs, and external references to the new owner. - Commit the final runbook with exact discovered resource names/commands, ordering, permissions, verification evidence, and recovery instructions. Remove dependence on old-URL redirects; clearly identify historical references that intentionally remain. ## Rollback and stop conditions - Stop and keep automation paused on fetch/authentication failures, unexpected pruning, missing applications, or failed critical health checks. Existing workloads must not be deleted to work around a repository-access problem. - Document whether reverse transfer to `martin` is permitted and whether the old namespace remains available before cutover. If reversing transfer, restore working repository access first, then revert canonical URLs/allowlists/credentials and regenerate manifests from the known-good Git state. - If reverse transfer is unavailable, recover repository access at `Personal/cluster` and roll back configuration there; do not restore a source URL that no longer hosts an accessible repository. - Document secure recovery of bootstrap/root source settings and the exact restoration of paused automation. Temporary live recovery changes must be reconciled back into Git. Re-verify all environments before resuming automation. ## Acceptance criteria - [ ] A reviewed runbook describes preparation, ordered cutover, verification, stop conditions, and rollback, including the self-managed Argo CD bootstrap dependency. - [ ] The existing repository is transferred to `Personal/cluster` with history, issues, PRs, permissions, and protections verified. - [ ] Nix source and generated manifests for all affected environments use the canonical new URL; no generated YAML or encrypted SOPS values were hand-edited. - [ ] Argo CD reconciles all environments from the new repository with no unexpected pruning or migration-caused downtime; original automation settings are restored. - [ ] Woodpecker, Renovate, webhooks, status checks, and required fetch/write credentials work under the new organization. - [ ] Container packages and image pulls remain functional; registry relocation is not silently bundled into this task. - [ ] Documentation, bootstrap instructions, and local-remotes guidance are updated, and verification evidence plus the tested/validated recovery plan are recorded.
Author
Owner

Phase 1 (prep) complete — awaiting signed commit + transfer window

Prep changes are ready in the working tree for a local branch (migration/personal-org), not yet committed (GPG signing is done by @martin).

Changes

  • modules/common.nix: networking.clusterRepo → https://code.${domain.root}/Personal/cluster.git. networking.registry deliberately unchanged (no app references it; images already publish to code.martials.no/personal/<app>).
  • .woodpecker/switch.yml: automerge endpoint now /api/v1/repos/${CI_REPO}/pulls/... instead of the hardcoded martin/cluster path.
  • AGENTS.md: issue-tracking link/slug → Personal/cluster; new "Canonical Repository" subsection with the GitOps URL, the local SSH remote, git remote set-url guidance, and the note that image pulls are independent of repo ownership.
  • Generated manifests regenerated via just switch {dev,infra,prod} — 11 Application repoURL lines + manifests/dev/argocd/Secret-argocd-repo-cluster.yaml url (decodes to https://code.martials.no/Personal/cluster.git). No other manifest churn: no deletions, no namespace/identity changes.

Validation

  • just fmt — clean (0 files changed).
  • just test — 108/108.
  • just test-IT — 11/11.
  • just diff {dev,infra,prod} reviewed before switching: only repoURL/url hunks.
  • git grep martin/cluster over tracked files — no remaining references.

Cluster inventory (verified live, not just Git)

  • 12 Argo CD Applications, all Synced/Healthy, all sourcing https://code.martials.no/martin/cluster.git: apps-dev, apps-infra, apps-prod (roots, bootstrapped outside Git), argocd-dev (self-managed), plus 8 child apps.
  • Only one AppProject (default) with sourceRepos: ['*'] — no allowlist change required.
  • argocd-repo-cluster Secret carries only url + project; the repository is public, so Argo CD fetches anonymously and no credentials need rotating.
  • Woodpecker repo id 2 (forge_remote_id: 17); server runs with WOODPECKER_ORGS=Personal, so the new location matches existing policy.
  • Renovate already autodiscovers martin/* and Personal/* (autodiscoverNamespaces: ["Personal"]) — no config change expected.
  • Global Woodpecker pipelines (flake-check cron, pr-review, release) do not hardcode the repository slug.
  • grafana-dashboards dashboard link and backstage sample catalog still reference the old slug (separate repos, follow-up commits).

Follow-ups

  • #80 — stale Getting Started docs (staging environment, layout) found during the inventory.

Next: signed commit + PR (CI validates the build), then the cutover window (snapshot, pause root syncPolicy, transfer, repoint live Applications, ordered sync, automation restore, verification).

## Phase 1 (prep) complete — awaiting signed commit + transfer window Prep changes are ready in the working tree for a local branch (`migration/personal-org`), not yet committed (GPG signing is done by @martin). ### Changes - `modules/common.nix`: `networking.clusterRepo` → `https://code.${domain.root}/Personal/cluster.git`. `networking.registry` deliberately unchanged (no app references it; images already publish to `code.martials.no/personal/<app>`). - `.woodpecker/switch.yml`: automerge endpoint now `/api/v1/repos/${CI_REPO}/pulls/...` instead of the hardcoded `martin/cluster` path. - `AGENTS.md`: issue-tracking link/slug → `Personal/cluster`; new "Canonical Repository" subsection with the GitOps URL, the local SSH remote, `git remote set-url` guidance, and the note that image pulls are independent of repo ownership. - Generated manifests regenerated via `just switch {dev,infra,prod}` — 11 `Application` `repoURL` lines + `manifests/dev/argocd/Secret-argocd-repo-cluster.yaml` `url` (decodes to `https://code.martials.no/Personal/cluster.git`). No other manifest churn: no deletions, no namespace/identity changes. ### Validation - `just fmt` — clean (0 files changed). - `just test` — 108/108. - `just test-IT` — 11/11. - `just diff {dev,infra,prod}` reviewed before switching: only repoURL/url hunks. - `git grep martin/cluster` over tracked files — no remaining references. ### Cluster inventory (verified live, not just Git) - 12 Argo CD Applications, all `Synced/Healthy`, all sourcing `https://code.martials.no/martin/cluster.git`: `apps-dev`, `apps-infra`, `apps-prod` (roots, bootstrapped outside Git), `argocd-dev` (self-managed), plus 8 child apps. - Only one AppProject (`default`) with `sourceRepos: ['*']` — no allowlist change required. - `argocd-repo-cluster` Secret carries only `url` + `project`; the repository is public, so Argo CD fetches anonymously and no credentials need rotating. - Woodpecker repo id `2` (`forge_remote_id: 17`); server runs with `WOODPECKER_ORGS=Personal`, so the new location matches existing policy. - Renovate already autodiscovers `martin/*` and `Personal/*` (`autodiscoverNamespaces: ["Personal"]`) — no config change expected. - Global Woodpecker pipelines (`flake-check` cron, `pr-review`, release) do not hardcode the repository slug. - `grafana-dashboards` dashboard link and `backstage` sample catalog still reference the old slug (separate repos, follow-up commits). ### Follow-ups - #80 — stale Getting Started docs (`staging` environment, layout) found during the inventory. Next: signed commit + PR (CI validates the build), then the cutover window (snapshot, pause root `syncPolicy`, transfer, repoint live Applications, ordered sync, automation restore, verification).
martin self-assigned this 2026-11-15 13:57:15 +00:00
martin added reference migration/personal-org 2026-11-15 14:46:22 +00:00
Author
Owner

Cutover executed — Personal/cluster is live

Owner transfer and merge were performed by @martin at ~20:13 UTC; the cluster was repointed immediately afterwards. No maintenance window pause was needed: no root syncPolicy was suspended, and validation showed zero pruning.

What was done

Step Result
Snapshot kubectl -n argocd get applications -o yaml → /tmp/argocd-apps-pre-repoint.yaml (5426 lines) before any change
Live repoint spec.source.repoURL patched on all 14 Applications → https://code.martials.no/Personal/cluster.git
Repo registration argocd-repo-cluster Secret url → new URL (also matches the regenerated manifest)
Local clone git remote set-url origin ssh://forgejo@nidaros.secure.martials.no/Personal/cluster.git; git fetch --prune
Merge PR #81 squash-merged as 53840c0
Post-merge reconcile d4bc291 (comment-only) rebase-merged; all apps refreshed onto it

Verification evidence

  • Old URL is dead for git: git ls-remote ssh://…/martin/cluster.git → Forgejo: Cannot find repository. repo-server logged one status code: 301 from the old HTTPS URL during the transition (GenerateManifest), so repointing was required rather than relying on the redirect. No such errors afterwards.
  • All 14 Applications Synced/Healthy at d4bc2913a55cdcf8ffaed6d5f74a280470b1fb51 from the new URL: apps-dev, apps-infra, apps-prod, argocd-dev, argo-rollouts-infra, prometheus-stack-infra, martials.no-dev, martials.no-prod, recurring-events-dev, recurring-events-prod, sops-secrets-operator-dev, toolbox-dev, toolbox-prod, traefik-dev.
  • No pruning: status.operationState.syncResult.pruned length is 0 for every application; resource counts unchanged (e.g. argocd-dev 50, prometheus-stack-infra 90, traefik-dev 33).
  • Self-management intact: argocd-dev reconciled itself from the new URL, and the resulting argocd-repo-cluster Secret keeps the new URL.
  • Repository metadata: Personal/cluster is public, main remains default, 13 open issues (12 + #80) and 3 open PRs (#58, #44, #42) carried over, all 12 repository labels preserved.
  • Woodpecker: repo id 2 (forge_remote_id: 17) re-registered as Personal/cluster, active, default branch main.
  • CI at the new owner (PR #82, rebase-merged, branch deleted): pipeline 345 success; commit statuses ci/woodpecker/pr/switch and ci/woodpecker/pr/pr-review both success; branch head unchanged afterwards, i.e. nixidy switch produced no manifest drift.
  • Renovate: manual systemctl start renovate.service → autodiscovery reports 17 repositories including Personal/cluster; no config change needed because discovery already covered Personal/*.

Residual items

  1. Bot manifest write-back is not yet exercised. Pipeline 345 ran git add manifests/ with nothing to commit, so clank-bot's push permission at Personal/cluster is unproven. It will be exercised by the first manifest-affecting change; the criterion stays open until then.
  2. No branch protection rule exists on main — checked before (martin/cluster) and after (Personal/cluster): get_branch_protection returns "target couldn't be found" in both cases. Nothing was lost, but there is no protection or required status-check gate today.
  3. External references still name the old owner: grafana-dashboards (dashboards/cluster-overview.json link panel) and backstage (examples/forgejo-repos.yaml sample catalog) — separate repositories.
  4. Signature note: #81's squash merge produced an unsigned commit (53840c0, G? = U); #82 used rebase and preserved the signed commit (d4bc291, G), though Forgejo reports "No principal matched" until the SSH signing key is registered on the profile.

Documentation

  • AGENTS.md gained the "Canonical Repository" section (GitOps URL, local SSH remote, git remote set-url, image-ownership note) and the issue-tracking link now points at Personal/cluster — merged in #81.
  • Follow-up #80 covers the stale Getting Started docs found during the inventory.
## Cutover executed — `Personal/cluster` is live Owner transfer and merge were performed by @martin at ~20:13 UTC; the cluster was repointed immediately afterwards. No maintenance window pause was needed: no root `syncPolicy` was suspended, and validation showed zero pruning. ### What was done | Step | Result | |---|---| | Snapshot | `kubectl -n argocd get applications -o yaml` → `/tmp/argocd-apps-pre-repoint.yaml` (5426 lines) before any change | | Live repoint | `spec.source.repoURL` patched on all 14 Applications → `https://code.martials.no/Personal/cluster.git` | | Repo registration | `argocd-repo-cluster` Secret `url` → new URL (also matches the regenerated manifest) | | Local clone | `git remote set-url origin ssh://forgejo@nidaros.secure.martials.no/Personal/cluster.git`; `git fetch --prune` | | Merge | PR #81 squash-merged as `53840c0` | | Post-merge reconcile | `d4bc291` (comment-only) rebase-merged; all apps refreshed onto it | ### Verification evidence - **Old URL is dead for git**: `git ls-remote ssh://…/martin/cluster.git` → `Forgejo: Cannot find repository`. repo-server logged one `status code: 301` from the old HTTPS URL during the transition (`GenerateManifest`), so repointing was required rather than relying on the redirect. No such errors afterwards. - **All 14 Applications `Synced/Healthy`** at `d4bc2913a55cdcf8ffaed6d5f74a280470b1fb51` from the new URL: `apps-dev`, `apps-infra`, `apps-prod`, `argocd-dev`, `argo-rollouts-infra`, `prometheus-stack-infra`, `martials.no-dev`, `martials.no-prod`, `recurring-events-dev`, `recurring-events-prod`, `sops-secrets-operator-dev`, `toolbox-dev`, `toolbox-prod`, `traefik-dev`. - **No pruning**: `status.operationState.syncResult.pruned` length is 0 for every application; resource counts unchanged (e.g. `argocd-dev` 50, `prometheus-stack-infra` 90, `traefik-dev` 33). - **Self-management intact**: `argocd-dev` reconciled itself from the new URL, and the resulting `argocd-repo-cluster` Secret keeps the new URL. - **Repository metadata**: `Personal/cluster` is public, `main` remains default, 13 open issues (12 + #80) and 3 open PRs (#58, #44, #42) carried over, all 12 repository labels preserved. - **Woodpecker**: repo id `2` (`forge_remote_id: 17`) re-registered as `Personal/cluster`, active, default branch `main`. - **CI at the new owner** (PR #82, rebase-merged, branch deleted): pipeline `345` success; commit statuses `ci/woodpecker/pr/switch` and `ci/woodpecker/pr/pr-review` both `success`; branch head unchanged afterwards, i.e. `nixidy switch` produced no manifest drift. - **Renovate**: manual `systemctl start renovate.service` → autodiscovery reports 17 repositories including `Personal/cluster`; no config change needed because discovery already covered `Personal/*`. ### Residual items 1. **Bot manifest write-back is not yet exercised.** Pipeline `345` ran `git add manifests/` with nothing to commit, so `clank-bot`'s push permission at `Personal/cluster` is unproven. It will be exercised by the first manifest-affecting change; the criterion stays open until then. 2. **No branch protection rule exists on `main`** — checked before (`martin/cluster`) and after (`Personal/cluster`): `get_branch_protection` returns "target couldn't be found" in both cases. Nothing was lost, but there is no protection or required status-check gate today. 3. **External references still name the old owner**: `grafana-dashboards` (`dashboards/cluster-overview.json` link panel) and `backstage` (`examples/forgejo-repos.yaml` sample catalog) — separate repositories. 4. **Signature note**: #81's squash merge produced an unsigned commit (`53840c0`, `G?` = `U`); #82 used rebase and preserved the signed commit (`d4bc291`, `G`), though Forgejo reports "No principal matched" until the SSH signing key is registered on the profile. ### Documentation - `AGENTS.md` gained the "Canonical Repository" section (GitOps URL, local SSH remote, `git remote set-url`, image-ownership note) and the issue-tracking link now points at `Personal/cluster` — merged in #81. - Follow-up #80 covers the stale Getting Started docs found during the inventory.
Author
Owner

Closed — migration complete

Owner transfer, cutover, and verification are finished. Summary of the outcome:

  • Personal/cluster is the canonical GitOps repository; martin/cluster no longer exists and its SSH path does not resolve.
  • All 14 Argo CD Applications reconcile from https://code.martials.no/Personal/cluster.git, Synced/Healthy, with zero pruning across the move.
  • networking.clusterRepo and the generated manifests point at the new owner; AGENTS.md documents the canonical GitOps URL and local remote.
  • Woodpecker re-registered as Personal/cluster (repo id 2); PR pipeline, webhooks, and status checks verified at the new owner.
  • Renovate autodiscovery includes Personal/cluster.
  • History, branches, 13 open issues, 3 open PRs, and all repository labels carried over.

Known exceptions accepted at closure:

  • Bot manifest write-back was never exercised (no manifest drift occurred after the move). First manifest-affecting change will prove it; if it fails, re-check clank-bot push access at Personal/cluster.
  • Branch protection on main: none existed before or after the transfer. Explicitly out of scope per @martin.
  • External references: grafana-dashboards updated in Personal/grafana-dashboards#9 (open, awaiting review/merge; no gcx resources push run). backstage catalog refreshes automatically from the forge, so no change is needed.
  • Signatures: #81 landed unsigned via squash merge; subsequent merges use rebase and keep signed commits. Registering the SSH key as a Forgejo signing key would make them show as verified.

Follow-up left open: #80 (stale Getting Started docs).

## Closed — migration complete Owner transfer, cutover, and verification are finished. Summary of the outcome: - `Personal/cluster` is the canonical GitOps repository; `martin/cluster` no longer exists and its SSH path does not resolve. - All 14 Argo CD Applications reconcile from `https://code.martials.no/Personal/cluster.git`, `Synced/Healthy`, with zero pruning across the move. - `networking.clusterRepo` and the generated manifests point at the new owner; `AGENTS.md` documents the canonical GitOps URL and local remote. - Woodpecker re-registered as `Personal/cluster` (repo id `2`); PR pipeline, webhooks, and status checks verified at the new owner. - Renovate autodiscovery includes `Personal/cluster`. - History, branches, 13 open issues, 3 open PRs, and all repository labels carried over. Known exceptions accepted at closure: - **Bot manifest write-back** was never exercised (no manifest drift occurred after the move). First manifest-affecting change will prove it; if it fails, re-check `clank-bot` push access at `Personal/cluster`. - **Branch protection on `main`**: none existed before or after the transfer. Explicitly out of scope per @martin. - **External references**: `grafana-dashboards` updated in Personal/grafana-dashboards#9 (open, awaiting review/merge; no `gcx resources push` run). `backstage` catalog refreshes automatically from the forge, so no change is needed. - **Signatures**: #81 landed unsigned via squash merge; subsequent merges use rebase and keep signed commits. Registering the SSH key as a Forgejo signing key would make them show as verified. Follow-up left open: #80 (stale Getting Started docs).
Author
Owner

Correction to the signature note above

My earlier statement that #81's squash merge landed "unsigned" was wrong. 53840c0 is signed and Forgejo verifies it:

$ curl -s https://code.martials.no/api/v1/repos/Personal/cluster/git/commits/53840c0 | jq '.verification | {verified, reason, signer}'
{
  "verified": true,
  "reason": "Forgejo Instance / SHA256:/en+e39PkqipWawBfJfUkbgr9Sk8wlCHJk0vBEcLVIo",
  "signer": { "name": "Forgejo Instance", "email": "forgejo@martials.no" }
}

And d4bc291 verifies as the user key (martin / SHA256:4p1Mo…). The U I saw comes from local git, not from Forgejo: ~/.config/git/allowed_signers only lists host SSH keys under the git@martials.no principal, so instance-signed merges report No principal matched.

The instance already signs merges via [repository.signing] (FORMAT = ssh, MERGES = always, SIGNING_KEY = /run/forgejo/id_ed25519.pub, SIGNING_NAME = Forgejo Instance), so no server-side key needs adding. The remaining work is local trust of that key, tracked in Personal/nixos-configuration#64.

## Correction to the signature note above My earlier statement that #81's squash merge landed "unsigned" was wrong. `53840c0` **is** signed and Forgejo verifies it: ```console $ curl -s https://code.martials.no/api/v1/repos/Personal/cluster/git/commits/53840c0 | jq '.verification | {verified, reason, signer}' { "verified": true, "reason": "Forgejo Instance / SHA256:/en+e39PkqipWawBfJfUkbgr9Sk8wlCHJk0vBEcLVIo", "signer": { "name": "Forgejo Instance", "email": "forgejo@martials.no" } } ``` And `d4bc291` verifies as the user key (`martin / SHA256:4p1Mo…`). The `U` I saw comes from local `git`, not from Forgejo: `~/.config/git/allowed_signers` only lists host SSH keys under the `git@martials.no` principal, so instance-signed merges report `No principal matched`. The instance already signs merges via `[repository.signing]` (`FORMAT = ssh`, `MERGES = always`, `SIGNING_KEY = /run/forgejo/id_ed25519.pub`, `SIGNING_NAME = Forgejo Instance`), so no server-side key needs adding. The remaining work is local trust of that key, tracked in Personal/nixos-configuration#64.
Sign in to join this conversation.
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Reference
Personal/cluster#76
No description provided.