railiance-platform/workplans/RPF-WP-0044-argocd-phase-b-adopt-existing-applications.md

id	type	title	domain	repo	status	flavor	owner	topic_slug	created	updated	related	state_hub_workstream_id
RPF-WP-0044	workplan	ArgoCD phase B: adopt the four existing Applications on railiance01	financials	railiance-platform	blocked	planning	railiance-platform	railiance	2026-09-21	2026-10-11	RPF-WP-0043 RPF-WP-0022	98140775-3b9a-5cf9-9af6-722502d487dc

The founder, Bernd Worsch, decided ArchitectureBlueprint §5.6 on 2026-09-21, exercising GOVERN @ estate: option 1, adopt properly. Phase A installed Argo CD Core v3.5.3 on railiance01 (record: the-custodian/docs/kubernetes-change-gate-decision.md). This plan is phase B: bring the four Applications in argocd/applications/ under that ArgoCD, one at a time, each with its own founder go-ahead. Terms follow SecurityCanon Mode of Authority v0.2.0 (draft).

This plan changes nothing live. Every task that would change railiance01 is wait on the founder. Each live step is ADMINISTER @ realm:kubernetes/railiance01 (bootstrap) or CONSTRUCT @ railiance-platform with the merge as the gate, both activation=APPROVED.

Related, kept separate: RPF-WP-0043 onboards rapp-policy-nexus, which is not one of these four. It needs T02 of this plan (a working root app on railiance01) before its own T04. Phase C, retiring coulombcore's ArgoCD, is not planned here (T08).

Read-only evidence, 2026-09-21

Full record: docs/evidence/argocd-phase-b-adoption-diff-2026-09-21.json. Rendered locally (helm template, kubectl kustomize), then kubectl diff client-side and --server-side dry run over ssh railiance01. Nothing was applied.

App	Live spec diff	Would prune	Repo credential	Blocker
openbao-secretstore	none (1 ClusterSecretStore)	nothing	none, public	none
issue-core	none after repointing	nothing	new: rapp-issue-core is private	declared path issue-core.git k8s/railiance no longer exists
target-revenue	none on Deployment/Service/Ingress/ExternalSecret/Cluster	nothing	none, public	CNPG Cluster not in railiance-tenants whitelist; two Sync-hook Jobs
external-secrets	none (39 objects, 20 CRDs)	nothing now	none, public Helm repo	cluster-wide blast radius; CRDs must never be pruned

Nothing is pruned at adoption because no live object carries an ArgoCD tracking annotation; ArgoCD only prunes what it tracks. The one change every adoption makes is that tracking annotation, on metadata only, so no rollout.

Sync settings found: railiance-apps-root and all four children declare automated: {prune: true, selfHeal: true}; none has a finalizer. Applying argocd/bootstrap/ as committed would therefore adopt all four at once. T01 removes that before anything is applied.

Also found: the external-secrets Helm release is stuck at revision 7, pending-upgrade, since 2026-07-07 (values identical to the Application); issue-core is now server-side applied by rapp-issue-core; target-revenue runs a mutable image tag (0.1.4). ArgoCD on railiance01 has 0 AppProjects, 0 Applications, 0 repository Secrets, and there is no argocd CLI on the node, so the commands below use kubectl against the Application resource. Not verified: repo-server egress to Forgejo and charts.external-secrets.io; T02 proves it.

Neutralise automated sync and fix the declared state

id: RPF-WP-0044-T01
status: done
priority: high
state_hub_task_id: "2d7230f5-50c8-5eee-a050-1501ce5af62a"

Done 2026-09-21, commit c3ebd6d. The founder chose Option A (GOVERN @ estate): a railiance01-only source path. Repository-only; nothing applied. argocd/applications/ and argocd/bootstrap/ are untouched, because coulombcore's ArgoCD still reads argocd/applications/ on main and its bootstrap may be re-applied from argocd/bootstrap/. coulombcore could not be read in this session, so the railiance01 bootstrap has its own path.

Declared (all under argocd/railiance01/):

1. bootstrap/: the three AppProjects and railiance-apps-root with no automated block, path argocd/railiance01/applications, targetRevision: main (each sync pins its revision in the operation).
2. bootstrap/01-railiance-tenants-project.yaml: postgresql.cnpg.io/Cluster added to namespaceResourceWhitelist. The other two AppProjects are copies of argocd/bootstrap/.
3. applications/ holds only README.md. A placeholder is needed: git does not keep an empty directory and a missing path is a ComparisonError. ArgoCD's directory source reads only *.yaml/*.yml/*.json, so it renders zero children (confirmed live by T02).
4. drafts/: the four children, no automated, no finalizer, pinned: openbao-secretstore d2dbc19 (railiance-platform), issue-core a78f38c (rapp-issue-core.git, path manifests), target-revenue f1109d5, external-secrets chart 0.16.1. No root syncs drafts/.
5. argocd/repositories/rapp-issue-core.repository.sops.yaml.template (no value); issue-core.repository.sops.yaml.template marked obsolete.

Offline proof: kubectl kustomize argocd/railiance01/bootstrap renders exactly 3 AppProjects and 1 Application with no automated field; ssh railiance01 'kubectl apply --dry-run=server -f -' of that render returned rc=0 (all four "created (server dry run)").

Hazard kept as is: make argocd-bootstrap-deploy still defaults to argocd/bootstrap (automated root). Do not run it on railiance01.

Apply the AppProjects and the root app, automated sync off (live)

id: RPF-WP-0044-T02
status: done
priority: high
state_hub_task_id: "7d97ee9c-b2fb-520f-8468-3ecf7eea6c01"

Done 2026-09-21 (founder go-ahead; custodian session). 3 AppProjects + railiance-apps-root applied from the clean export of c3ebd6d; root Synced/Healthy, op Succeeded at c3ebd6d, zero children, 0 ArgoCD-managed namespaces. Repo-server reaches Forgejo. One OrphanedResourceWarning (project-level, informational).

Applies the render of argocd/railiance01/bootstrap at commit c3ebd6dddc830373c75144da178f829b47c3e5de, from a clean export (not the working tree), then syncs the root by hand at that commit.

cd ~/railiance-platform && git fetch origin && git merge-base --is-ancestor c3ebd6dddc830373c75144da178f829b47c3e5de origin/main && echo pinned-commit-on-main
R=$(mktemp -d) && git -C ~/railiance-platform archive c3ebd6dddc830373c75144da178f829b47c3e5de argocd/railiance01/bootstrap | tar -x -C "$R"
kubectl kustomize "$R/argocd/railiance01/bootstrap" > "$R/bootstrap.yaml" && grep -c '^kind: AppProject' "$R/bootstrap.yaml" && grep -c automated "$R/bootstrap.yaml"   # expect 3 and 0
ssh railiance01 'kubectl apply --dry-run=server -f -' < "$R/bootstrap.yaml"
ssh railiance01 'kubectl apply -f -' < "$R/bootstrap.yaml"
ssh railiance01 'kubectl -n argocd patch application railiance-apps-root --type merge -p "{\"operation\":{\"initiatedBy\":{\"username\":\"founder-approved\"},\"sync\":{\"revision\":\"c3ebd6dddc830373c75144da178f829b47c3e5de\",\"prune\":false}}}"'
ssh railiance01 'kubectl -n argocd get appprojects,applications'
ssh railiance01 'kubectl -n argocd get application railiance-apps-root -o jsonpath="{.status.sync.status} {.status.operationState.phase} {.status.operationState.syncResult.revision} resources={.status.resources}{\"\n\"}"'

Expect three AppProjects and railiance-apps-root Synced, operation Succeeded at c3ebd6d, and no resources (zero children). A ComparisonError means the repo-server cannot reach Forgejo: stop and fix egress before any app task. No repository Secret is needed; coulomb/railiance-platform is public.

Rollback: kubectl -n argocd delete application railiance-apps-root (no finalizer, no children), then delete the three AppProjects. Nothing else is touched.

Per-app procedure (T03–T06)

Every adoption task follows the same steps. <app> and <sha> vary.

1. Diff. Re-run the evidence render and kubectl diff --server-side at the commit or chart version being pinned. Any spec change, any missing whitelist kind, or a different image: stop.
2. Merge the child Application (no automated) from argocd/railiance01/drafts/ into argocd/railiance01/applications/, then sync the root by hand so it creates the child object only:

   ssh railiance01 'kubectl -n argocd patch application railiance-apps-root --type merge -p "{\"operation\":{\"initiatedBy\":{\"username\":\"founder-approved\"},\"sync\":{\"revision\":\"<sha>\",\"prune\":false}}}"'
   

3. ArgoCD's own diff: read kubectl -n argocd get application <app> -o jsonpath='{.status.sync.status} {.status.health.status}' and .status.resources. Only metadata differences are acceptable.
4. Manual sync, prune off, hooks skipped (apply strategy):

   ssh railiance01 'kubectl -n argocd patch application <app> --type merge -p "{\"operation\":{\"initiatedBy\":{\"username\":\"founder-approved\"},\"sync\":{\"revision\":\"<sha>\",\"prune\":false,\"syncStrategy\":{\"apply\":{}}}}}"'
   

   Then verify Synced/Healthy and run the owner's live check.
5. Prove it: Synced and Healthy for at least 24 hours (7 days for external-secrets). Only then, with a second go-ahead, a merge adds automated: {selfHeal: true}; prune: true is a third step once the app's tracked set is confirmed complete.
6. Record docs/evidence/<date>-<app>-argocd-adoption.json.

Rollback:

• R1, before automated sync: revert the merge, then kubectl -n argocd delete application <app>. No finalizer, so nothing cascades; the workload keeps running. Tracking annotations stay and are harmless.
• R2, after automated sync: revert the automated-sync commit first and sync the root by hand, then R1. Never hand-patch a child while the root self-heals; that is BREAK_GLASS, recorded and reconciled into git.
• Workload restore: only once ArgoCD no longer manages the app, with the previous deploy method named in the task.

Adopt openbao-secretstore (lowest risk)

id: RPF-WP-0044-T03
status: done
decision_id: 89f1058f-33bc-47a8-8a49-0b44d426c67e
priority: high
state_hub_task_id: "6d5fc801-e361-579b-bcd8-6fe719a82e94"

2026-10-11: requalified after founder decision D2 (ArgoCD owns production) resolved — see the-custodian docs/decisions/2026-10-11-security-unblock-decisions.md.

Adopted 2026-09-21 (founder go-ahead; custodian session): diff at d2dbc19 rc=0; child merged in 182e788; root synced at 182e788; child manual sync (apply strategy, prune off) Succeeded at d2dbc19, Synced/Healthy, automated off. ClusterSecretStore/openbao Valid with tracking annotation; issue-core-runtime SecretSynced. Status progress = in the 24h proving period; selfHeal needs a second go-ahead. Evidence: docs/evidence/2026-09-21-openbao-secretstore-argocd-adoption.json.

One object, ClusterSecretStore/openbao, zero diff, public source. It is the store issue-core's ExternalSecret reads, so adopt it before issue-core. The 24 other ClusterSecretStores on railiance01 are not in the kustomization and stay untracked (declared gap, RPF-WP-0043-T05). Live check: kubectl get clustersecretstore openbao stays Valid, and externalsecret/issue-core-runtime stays SecretSynced. Workload restore: kubectl apply -f argocd/platform-addons/openbao-secretstore/openbao.clustersecretstore.yaml.

Exact commands (the root sync pins the merge commit $SHA; the child is pinned to d2dbc19, the last commit touching its source path):

# 1. Diff at the pinned source commit (expect no output, rc=0)
cd ~/railiance-platform && git fetch origin && git checkout main && git pull --ff-only
R=$(mktemp -d) && git archive d2dbc19c254247652c49fda8721c80d53bca206a argocd/platform-addons/openbao-secretstore | tar -x -C "$R"
ssh railiance01 'kubectl diff --server-side -f -' < <(kubectl kustomize "$R/argocd/platform-addons/openbao-secretstore"); echo rc=$?
# 2. Merge the child (drop the 3-line DRAFT header), push, sync the root at that commit
sed '1,3d' argocd/railiance01/drafts/openbao-secretstore.application.yaml > argocd/railiance01/applications/openbao-secretstore.application.yaml
git rm -q argocd/railiance01/drafts/openbao-secretstore.application.yaml
git add argocd/railiance01/applications/openbao-secretstore.application.yaml
git commit -m "Adopt openbao-secretstore on railiance01 (RPF-WP-0044-T03)" && git push origin main
SHA=$(git rev-parse HEAD) && echo "$SHA"
ssh railiance01 "kubectl -n argocd patch application railiance-apps-root --type merge -p '{\"operation\":{\"initiatedBy\":{\"username\":\"founder-approved\"},\"sync\":{\"revision\":\"$SHA\",\"prune\":false}}}'"
# 3. ArgoCD's own diff (expect OutOfSync or Synced, one resource ClusterSecretStore/openbao, metadata-only)
ssh railiance01 'kubectl -n argocd get application openbao-secretstore -o jsonpath="{.status.sync.status} {.status.health.status} {.status.resources}{\"\n\"}"'
# 4. Manual sync, prune off, apply strategy (hooks skipped)
ssh railiance01 'kubectl -n argocd patch application openbao-secretstore --type merge -p "{\"operation\":{\"initiatedBy\":{\"username\":\"founder-approved\"},\"sync\":{\"revision\":\"d2dbc19c254247652c49fda8721c80d53bca206a\",\"prune\":false,\"syncStrategy\":{\"apply\":{}}}}}"'
ssh railiance01 'kubectl -n argocd get application openbao-secretstore -o jsonpath="{.status.sync.status} {.status.health.status} {.status.operationState.phase}{\"\n\"}"'
ssh railiance01 'kubectl get clustersecretstore openbao; kubectl -n issue-core get externalsecret issue-core-runtime'

Rollback (R1): git revert the merge commit and push, then ssh railiance01 'kubectl -n argocd delete application openbao-secretstore' (no finalizer; the ClusterSecretStore keeps running).

2026-10-03: human gate routed to hub decision 89f1058f-33bc-47a8-8a49-0b44d426c67e (D2: ArgoCD owns production: go-ahead sequence for railiance-apps-root, issue-core, policy-nexus and activity-core automation) by CUST-WP-0075; resolving that decision raises C-40 here.

2026-10-11: 24h adoption proof verified; T03 done. scripts/argocd_adoption_proof.py freshly verifies estate IP 92.205.62.239, local-cluster destination, exact source pin d2dbc19, one tracked ClusterSecretStore/openbao, manual idle policy, successful Sep21 apply/prune-off sync, exact ownership annotation and uninterrupted Ready conditions on the store and issue-core consumer since Aug19. The completed proving period is 1,735,984 seconds. Current Application is Synced/Healthy; the nonblocking OrphanedResourceWarning is retained and no broader namespace adoption is claimed.

Both server-side and client-side native pinned-render diffs return 0 after canonical Argo v3 annotation injection. This is bound to the observed v3.5.3 controller, argocd-cm tracking configuration and the Application destination namespace. The cluster-scoped store's tracking ID uses external-secrets as Argo's destination-namespace fallback; no instance label is injected in annotation mode. Same-source-API-version server dry-runs show no changed JSON paths. Only source-derived desired fields and Argo's exact tracking annotation are rendered; no live values are copied into the desired source, and unknown or substantive diffs fail closed. Upstream implementation: https://github.com/argoproj/argo-cd/blob/v3.5.3/util/argo/resource_tracking.go (UnstructuredToAppInstanceValue, SetAppInstance).

Evidence: docs/evidence/2026-10-11-openbao-adoption-proof/: observation-argo-normalized.json, proof-argo-normalized-final.json, and validation.json. The proof binds the verifier and observation digests. Thirteen meaningful fixture tests and compilation pass. Earlier empty-namespace and raw source diff refusals remain visible: raw source omits Argo-owned tracking and last-applied metadata; an initial API-version comparison also used the API default rather than the pinned source version. No native mutation, Secret read or annotation value/error output occurred. This closes the existing adoption's proving period without changing sync/automation policy or using a new human gate.

Adopt issue-core (from rapp-issue-core)

id: RPF-WP-0044-T04
status: done
intervention_note: "Orange attended Platform OpenBao OIDC/MFA login for the already D2-approved exact repository reader; no second source-scope approval."
decision_id: 89f1058f-33bc-47a8-8a49-0b44d426c67e
priority: high
state_hub_task_id: "0e619658-0453-51ed-b9a0-bed28b5544fd"

D2 already authorizes this exact private repository read-only reader and the ordered adoption. No renewed founder go-ahead is required. The remaining human act is the existing Platform OpenBao attended OIDC/MFA login; route it through the single attended-window checklist.

Source preparation now disables wrapper direct deployment/rollback and verifies Argo ownership using the exact HTTPS or SSH source URL. The fixed credential executor is scripts/provision_argocd_issue_core_repository.py, reviewed in docs/evidence/2026-10-11-issue-core-credential/owner-review.json. It creates one repository-only read-only SSH deploy key, stores it at platform/operators/argocd/repositories/rapp-issue-core, response-wraps for 60 seconds, unwraps once, proves consumed-token denial, and creates only argocd/rapp-issue-core-repository. Tests and installed Bao synthetic checks pass; native issuance/adoption have not run. The contract example is incomplete and cannot authorize execution.

Public host trust and exact project admission are prepared in owner source and the bounded scripts/argocd_public_transport.py; its server dry-run passed. Preserve each exact attempt and any partial failure. Finish those prerequisites, seal the owner-reviewed contract, use the attended login, then pin the newly committed wrapper source in the Application, inspect client-side native diff, manually sync with prune/force disabled, and run the bounded owner verifier. Image releases and rollback use Git revisions plus Argo; direct make rollback is disabled. Never adopt the stale wrapper revision that predates the ExternalSecret metadata guard.

2026-10-20 (CUST-WP-0079-T06 attended-window prep, agent): prerequisites done. Host trust and exact project admission were executed 2026-10-11 (the-custodian docs/evidence/2026-10-11-issue-core-public-transport/execution.json); a fresh dry-run refused with target_preimage_changed (already applied, kept as docs/evidence/2026-10-11-issue-core-transport/2026-10-20-dry-run-r1.json) and --action verify passed (2026-10-20-verify-r1.json). ssh-keyscan from railiance01 matches the pinned provider key. Sealed contract docs/evidence/2026-10-11-issue-core-credential/contract-sealed-2026-10-20.json (sha256 1ce9d2fdc39395407a01e358bfd0179f83f35c0a448e6658eb8f5aab351d3d3a, expires 2026-10-27) passes the executor's validate_contract; 17 executor/transport tests pass. Wrapper pin target is rapp-issue-core 186c817b67d3919a64acdddbd3889e539841f946. The only remaining step is the founder's attended login running the executor; exact commands and the agent follow-on (pin, diff, manual sync, verifier) are in docs/runbooks/rpf-wp-0044-t04-attended.md. No secret, credential or OpenBao call was made.

2026-10-20 (adoption, agent, founder explicit go + D2): the journal contract_sha256 0fd6d6fa... is sha256(json.dumps(contract, sort_keys=True)) of the sealed contract (recomputed, exact match; the raw file hash 1ce9d2fd... was enforced at launch). Child placed in argocd/railiance01/applications/issue-core.application.yaml (SSH repoURL, 186c817) in d70422c and applied directly, because railiance-apps-root is also OutOfSync on activity-core, informed-decision-requester and state-hub-memo-dispatch (make argocd-pin-drift: all ok). Argo fetched the SSH source at 186c817 without ComparisonError; Healthy, OutOfSync (8 untracked objects). Client-side diff of the pinned render (T05 method): rc=0; the render holds no credential objects. Manual sync not performed: the agent harness auto-mode classifier denied the Application sync-operation patch. Nothing was rolled back because no workload changed; the new Application has no automated sync. Remaining: the founder runs (or permits) runbook step 4 sync, then the step 5 verifier. Receipt: docs/evidence/2026-10-20-issue-core-argocd-adoption.json.

2026-10-20 (attended window, CUST-WP-0079-T06): founder completed the Platform OpenBao OIDC/MFA login and the fixed executor at 10:37Z. Journal docs/evidence/2026-10-11-issue-core-credential/execution-journal-r1.json: state delivered_not_adopted, consumed-wrap replay denied (400), no secret values observed by the agent, custody version 1, destination argocd/rapp-issue-core-repository. The human gate is cleared. Remaining agent steps per docs/runbooks/rpf-wp-0044-t04-attended.md: pin the Application, client-side diff, manual sync (prune/force off), bounded verifier. The journal's contract_sha256 (0fd6d6fa…) differs from the sealed file's sha256 (1ce9d2fd…). Reconcile this before the adoption sync.

2026-10-20 adoption complete. The founder ran the manual sync (apply strategy, prune/force off). Argo reports issue-core Synced/Healthy at 186c817, operation Succeeded. The bounded verifier make verify-live passed with all checks true: healthz 200, unauthenticated ingest 401, immutable image, ExternalSecret Ready, no secret values observed. Receipt: docs/evidence/2026-10-20-issue-core-verify-live.json. The runbook's ssh quoting was fixed. Automated sync is still off; enabling it is a separate decision. T04 done.

Adopt target-revenue

id: RPF-WP-0044-T05
status: done
decision_id: 89f1058f-33bc-47a8-8a49-0b44d426c67e
priority: medium
state_hub_task_id: "d418068a-6fb0-5416-aac5-d23c93924d9c"

2026-10-11: requalified after founder decision D2 (ArgoCD owns production) resolved — see the-custodian docs/decisions/2026-10-11-security-unblock-decisions.md.

Adopted 2026-09-21 (founder go-ahead; custodian session). The server-side diff at f1109d5 conflicted on resources.limits.cpu (live 1 owned by kubectl-client-side-apply vs declared 1000m: the same quantity); the client-side diff, which is what the apply strategy uses, was rc=0 for all five non-hook objects. Child merged in 6eeb826; manual sync with the apply strategy, prune off: Succeeded, Synced/Healthy, 0 Jobs created, same pod (36d, no restart), CNPG healthy 1/1, ExternalSecret SecretSynced, https://revenue.coulomb.social/healthz 200 (the live check path is /healthz, not /health). Status progress = proving period. Automated sync still needs target-revenue's confirmation on the two hook Jobs (asked 2026-09-21). Evidence: docs/evidence/2026-09-21-target-revenue-argocd-adoption.json.

Original preconditions: T01 item 2 (CNPG Cluster whitelisted) applied to the AppProject by hand, and target-revenue confirms the two Sync-hook Jobs (target-revenue-migrate, target-revenue-bootstrap-binky) are safe to re-run against production. They are absent live and would run on every sync with the default hook strategy. The adoption sync uses the apply strategy, which skips them. Once automated sync is enabled they run on each sync, so the second go-ahead depends on that confirmation. Ask target-revenue to pin an image digest instead of 0.1.4. Live check: https://revenue.coulomb.social health and the CNPG cluster healthy. Workload restore: kubectl apply -k k8s/railiance from the previous target-revenue commit.

2026-10-03: human gate routed to hub decision 89f1058f-33bc-47a8-8a49-0b44d426c67e (D2: ArgoCD owns production: go-ahead sequence for railiance-apps-root, issue-core, policy-nexus and activity-core automation) by CUST-WP-0075; resolving that decision raises C-40 here.

2026-10-11: manual adoption proving complete. Fresh read-only proof at 2026-10-11T19:51:12.812423+00:00 verified exact pin a3a8a27a8cda32ae3c7b55353ee16a131c50a0a0, manual apply strategy, prune/force off and no pending operation. The last successful pinned apply finished 2026-09-28T14:23:27Z; 1142865 seconds of proving exceeds 24 hours. All five non-hook objects are tracked and Synced; Deployment Available (generation observed), ExternalSecret Ready and CNPG Ready (1/1). ArgoCD remains Synced/Healthy; normalized native client-side diff is zero and https://revenue.coulomb.social/healthz returns 200. Both declared Sync-hook Jobs are absent and excluded from the apply-strategy diff. No live mutation or Secret read was performed. Receipt: docs/evidence/2026-10-11-target-revenue-adoption-proof/receipt.json; sanitized observation alongside it. Reusable check: scripts/workload_adoption_proof.py; 35 new failure/normalization tests plus 13 existing OpenBao adoption tests pass.

This closes T05 adoption/proving only. Child automated sync stays off. The existing target-revenue owner handoff for hook re-execution safety remains a prerequisite to future automation; it is not silently resolved and no new residual is invented here. The observed OrphanedResourceWarning is non-blocking; other Application condition types fail verification.

Adopt external-secrets (highest risk)

id: RPF-WP-0044-T06
status: wait
decision_id: 89f1058f-33bc-47a8-8a49-0b44d426c67e
priority: medium
depends_on: [RPF-WP-0044-T04, RPF-WP-0044-T05]
needs_human: true
blocking_reason: "Attended founder run of scripts/rpf_wp_0044_t06_attended.sh (KUBECTL_ROUTE=founder-sudo): fresh pre-checks, Application create, typed ADOPT, manual SSA sync prune/force off, post-checks. Runbook docs/runbooks/rpf-wp-0044-t06-eso-adoption.md."
state_hub_task_id: "4ca54868-c31f-5e3d-b7d3-a15b84c15f80"

2026-10-11: requalified after founder decision D2 (ArgoCD owns production) resolved — see the-custodian docs/decisions/2026-10-11-security-unblock-decisions.md.

Authorized by D2; waits on predecessor automated proof, not another founder go. The draft already carries CRD prune/delete protection (verified preparation 2026-09-27; 39 objects, 20 protected CRDs, metadata-only diff). That historical receipt is preparation, not current adoption proof. The deterministic scripts/eso_adoption_readiness.py checks the draft and bound predecessor receipts before a fresh native render/diff is considered. ESO serves every ExternalSecret on the cluster (25 ClusterSecretStores). Before the merge, add crds.annotations: {argocd.argoproj.io/sync-options: "Prune=false,Delete=false"} to the values and re-run the diff; pruning a CRD would delete every ExternalSecret and store. Adopt with ServerSideApply=true as declared. Watch for drift on caBundle fields injected by the cert-controller; if ArgoCD shows it, add ignoreDifferences before enabling self-heal. After adoption, retire the stale pending-upgrade Helm release record (its sh.helm.release.v1.external-secrets.v* Secrets) under a separate reviewed cleanup; ArgoCD does not track them. Live check: all ClusterSecretStores keep their current Valid state (two, openbao-activity-core and openbao-email-connect, are already ValidationFailed). Workload restore: revert the Platform Application to the previously reviewed chart 0.16.1 values in Git, then manually sync through ArgoCD; never prune/delete CRDs or silently replace the Helm release record as part of adoption.

2026-10-03: human gate routed to hub decision 89f1058f-33bc-47a8-8a49-0b44d426c67e (D2: ArgoCD owns production: go-ahead sequence for railiance-apps-root, issue-core, policy-nexus and activity-core automation) by CUST-WP-0075; resolving that decision raises C-40 here.

2026-10-21 (preparation, agent, no live mutation): scripts/eso_adoption_readiness.py passes the source check (chart 0.16.1, 20/20 CRDs Prune=false,Delete=false, ServerSideApply, manual, no finalizer, no Force/Replace) but reports T04 and T05 missing. The 2026-10-20 issue-core verify-live receipt has no authenticated smoke statuses (it ran without --smoke, and --smoke creates and closes a Forgejo issue, so it is not read-only) and is outside the 300 s window. A fresh read-only target-revenue proof failed because ssh railiance01 (tegwick) can no longer read the k3s kubeconfig after the CUST-WP-0073 cutover. Offline render of 0.16.1 with the draft values: 39 objects, no ClusterSecretStore or ExternalSecret rendered, webhooks declare no caBundle, the webhook-cert holder is metadata-only; no ignoreDifferences added yet. The fresh live client-side diff and the store baseline were not run (no agent cluster read path). Receipts: docs/evidence/2026-10-21-eso-adoption-prep/. Runbook with the exact founder sync command: docs/runbooks/rpf-wp-0044-t06-eso-adoption.md.

2026-10-21 (founder decisions, agent preparation): the founder will run ONE attended script from their own terminal through their admin path, and allowed the issue-core --smoke controlled write in that pre-check. Agent RBAC is not widened. New scripts/kube_route.py adds KUBECTL_ROUTE (agent default, founder-sudo = ssh railiance01 sudo -n …; no ssh -t, since sudo -n keeps pipes clean). It is used by the transport, adoption-proof, workload-proof, activity-readiness, pin-drift and service-observation scripts and by rapp-issue-core tools/verify_live.py. scripts/rpf_wp_0044_t06_attended.sh runs the fresh proofs, readiness gate, render, client-side diff, structured classification and store baseline, stops on a no-go, then creates the Application, requires typed ADOPT, syncs (prune/force off, SSA) and runs the post-checks and refresh sample; on failure it prints the rollback and never prunes. Evaluator scripts/eso_attended_checks.py; credential-kind objects are split out before any live read. Offline validation: shellcheck clean, --dry-run prints commands only, 118 tests pass in the affected files (new: test_kube_route, test_eso_attended_checks; rapp-issue-core: test_kube_route). Not verified live: sudo -n pipe behaviour under the host use_pty default; the --precheck-only run will show it.

Turn the root back to the production lane

id: RPF-WP-0044-T07
status: wait
decision_id: 89f1058f-33bc-47a8-8a49-0b44d426c67e
depends_on: [RPF-WP-0044-T03, RPF-WP-0044-T04, RPF-WP-0044-T05, RPF-WP-0044-T06]
blocking_reason: "Restore prune/selfHeal once T03-T06 have automated proof; no further founder go-ahead (D2)."
priority: medium
state_hub_task_id: "844ff115-def2-55ca-ab01-413ba7e6857d"

2026-10-11: requalified after founder decision D2 (ArgoCD owns production) resolved — see the-custodian docs/decisions/2026-10-11-security-unblock-decisions.md.

Waits on the founder's go-ahead. After T03–T06 are proven, restore automated: {prune: true, selfHeal: true} on railiance-apps-root, as the change-gate decision's production row expects. From then on a merge to the railiance01 path is a live production change, and RPF-WP-0043-T04 (policy-nexus) can use it.

2026-10-03: human gate routed to hub decision 89f1058f-33bc-47a8-8a49-0b44d426c67e (D2: ArgoCD owns production: go-ahead sequence for railiance-apps-root, issue-core, policy-nexus and activity-core automation) by CUST-WP-0075; resolving that decision raises C-40 here.

Plan phase C: retire coulombcore's ArgoCD

id: RPF-WP-0044-T08
status: done
priority: low
state_hub_task_id: "55d1382f-5862-5321-a1c9-96767764ba43"

2026-10-11: done by host-side read-only check (docs/evidence/2026-10-11-coulombcore-argocd-frozen.md). k3s certificates expired 2026-08-31, so the old ArgoCD cannot reconcile or prune. Founder: coulombcore is retired and the machine reused; retirement moves to NK-WP-0022-T08. Do not rotate certs or restart k3s there.

2026-10-11: requalified after founder decision D2 (ArgoCD owns production) resolved — see the-custodian docs/decisions/2026-10-11-security-unblock-decisions.md.

Planning only. Needs a read-only check of coulombcore's ArgoCD, outside this session's scope. Under Option A, retiring it also removes argocd/applications/. Also hand back to the cluster layer: the phase A install is not declared in any repository and its pods have no resource requests (BestEffort).

2026-10-03: human gate routed to hub decision 89f1058f-33bc-47a8-8a49-0b44d426c67e (D2: ArgoCD owns production: go-ahead sequence for railiance-apps-root, issue-core, policy-nexus and activity-core automation) by CUST-WP-0075; resolving that decision raises C-40 here.

Loose-end review — 2026-09-27

T03/T05 passed their elapsed healthy observation gate: current ArgoCD status is Synced/Healthy with last healthy transitions on September 21; openbao is Valid and issue-core-runtime is SecretSynced. They now wait on the remaining automation go-aheads and, for target-revenue, hook-owner acceptance. T06 local preparation is complete: the inactive ESO draft protects all 20 CRDs from prune/delete. The pinned 0.16.1 chart renders 39 objects; server-side diff is exactly the 20 protective metadata annotations, with no spec changes. All 25 ClusterSecretStores are Valid. T04 still needs the repository credential/production-owner contract; T07 depends on the per-app gates. T08 has a concrete retirement procedure at docs/argocd-coulombcore-retirement.md, but its required live inventory is blocked: SSH works and Kubernetes rejects both default and explicit local k3s kubeconfigs. No controller or workload was changed. Evidence: docs/evidence/2026-09-27-argocd-loose-end-status.json and docs/evidence/2026-09-27-eso-adoption-preparation.json.