Vault -> VPS secret pipeline: replaces manual kubectl/ssh secret copies

User correctly flagged this: manually copying secrets between home and
the VPS defeats the entire point of having Vault. Every VPS secret this
session (MinIO creds, Cloudflare token, CNPG certs, the pg-authentik-app
password) was a one-off kubectl create secret copy-paste, because the
VPS's k3s cluster had no Vault/ESO pipeline at all - this builds one.

- infrastructure/authentik/manifests/ha-postgres-app-pushsecret.yaml:
  home pushes pg-authentik-app's password into Vault at
  secret/vps/pg-authentik-app (PushSecret, not just Get - this password is
  CNPG-generated, not Vault-native, so it has to originate from a push).

- infrastructure/vps-eso/manifests/clustersecretstore.yaml: ESO on the
  VPS (installed via helm, out-of-band like k3s/cert-manager - see the
  file's own header) authenticates to home's Vault via AppRole (not
  Kubernetes auth - the VPS is a separate cluster with no federation to
  home's API server). Reachable via a new public
  https://vault.kube.huskypup.net record - Vault was deliberately kept
  off the public internet before this, explicitly confirmed with the user
  before opening it. Traffic goes through the same Istio ingress gateway
  already serving other public hosts, so unlike the ha-authentik-postgres
  NodePort case, no PeerAuthentication/AuthorizationPolicy change was
  needed - it arrives as a normal in-mesh call from the gateway's own
  identity, not raw external TCP to a pod.

  Found and fixed a real Vault gotcha while wiring this up: tried to
  remove the AppRole's token_bound_cidrs restriction (added first, before
  discovering Vault can't see the VPS's real source IP through the
  gateway - it only ever sees the gateway's own pod IP) by omitting the
  field from a follow-up  - that does NOT clear it, the
  AppRole role endpoint preserves omitted fields rather than resetting
  them to default. Had to explicitly write token_bound_cidrs=. Spent a
  while chasing a misleading 403 permission denied on auth/token/
  lookup-self before finding this - vault token capabilities said read
  was allowed (policy was fine), the actual rejection was IP-bound token
  use from an unbound context.

- infrastructure/vps-standby/authentik/manifests/
  pg-authentik-app-externalsecret.yaml: pulls it back down, Merge policy
  (only overwrites the password key - host/dbname/username stay local,
  CNPG still needs its own local -rw hostname for internal use).

- argocd-apps/vps-standby/vps-eso.yaml: new Application for the
  ClusterSecretStore.

Verified end-to-end: password now matches between home and VPS's
pg-authentik-app secrets via this pipeline (not the earlier manual patch),
confirmed by comparing both live secret values after ESO's sync.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

argocd-apps/vps-standby/vps-eso.yaml

@@ -0,0 +1,22 @@
+apiVersion: argoproj.io/v1alpha1
+kind: Application
+metadata:
+  name: vps-eso
+  namespace: argocd
+  annotations:
+    argocd.argoproj.io/sync-wave: "1"  # before any Application whose secrets it manages (authentik is wave 3)
+  finalizers:
+    - resources-finalizer.argocd.argoproj.io
+spec:
+  project: vps-standby
+  source:
+    repoURL: https://gitlab.kube.huskypup.net/Scooby/Homelabv4.git
+    targetRevision: main
+    path: infrastructure/vps-eso/manifests
+  destination:
+    name: vps-standby
+    namespace: external-secrets
+  syncPolicy:
+    automated:
+      prune: true
+      selfHeal: true

infrastructure/authentik/manifests/ha-postgres-app-pushsecret.yaml

@@ -0,0 +1,36 @@
+# Multi-site active failover pilot (see
+# /home/scooby/.claude/plans/jiggly-snacking-iverson.md) - closes the gap
+# the user correctly flagged: manually kubectl/ssh-copying secrets between
+# home and the VPS defeats the entire point of having Vault. This is the
+# first piece of a real Vault -> VPS pipeline (infrastructure/vps-eso/ has
+# the other half, ESO running on the VPS itself pulling this back down).
+#
+# pg-authentik-app is CNPG-generated, not Vault-native - each cluster
+# (home and VPS) independently generates its own random password for the
+# `app` role at bootstrap time. Since VPS's Postgres now REPLICATES from
+# home (WAL includes role/password changes), the two clusters' actual live
+# passwords are only in sync because of that replication - but the two
+# clusters' K8S SECRET OBJECTS never resync on their own (confirmed live
+# 2026-08-20: VPS's copy was stale). Pushing home's value into Vault, with
+# VPS's ESO pulling it back down on refreshInterval, makes this self-
+# healing instead of a manual one-time fix that goes stale again on the
+# next password rotation.
+apiVersion: external-secrets.io/v1alpha1
+kind: PushSecret
+metadata:
+  name: pg-authentik-app-to-vps
+  namespace: authentik
+spec:
+  refreshInterval: 5m
+  secretStoreRefs:
+    - name: vault-backend
+      kind: ClusterSecretStore
+  selector:
+    secret:
+      name: pg-authentik-app
+  data:
+    - match:
+        secretKey: password
+        remoteRef:
+          remoteKey: vps/pg-authentik-app
+          property: password

infrastructure/vps-eso/manifests/clustersecretstore.yaml

@@ -0,0 +1,84 @@
+# Multi-site active failover pilot (see
+# /home/scooby/.claude/plans/jiggly-snacking-iverson.md) - closes the gap
+# flagged 2026-08-20: manually kubectl/ssh-copying secrets between home
+# and the VPS defeats the point of having Vault at all. This gives the
+# VPS its own real Vault -> ESO pipeline instead.
+#
+# External Secrets Operator itself is installed directly via helm
+# (out-of-band, like k3s/cert-manager/Netbird - see
+# scripts/vps-bootstrap.sh's existing pattern for why those aren't
+# GitOps-managed either):
+#   helm repo add external-secrets https://charts.external-secrets.io
+#   helm install external-secrets external-secrets/external-secrets \
+#     --namespace external-secrets --create-namespace \
+#     --version 0.20.4 --set installCRDs=true
+# (0.20.4 matches home's version - see argocd-apps/infrastructure/
+# external-secrets.yaml)
+#
+# Auth: AppRole, not Kubernetes auth - home's existing vault-backend
+# ClusterSecretStore (infrastructure/vault/manifests/clustersecretstore.yaml)
+# uses Vault's kubernetes auth method, which verifies a ServiceAccount JWT
+# against THAT cluster's own API server - doesn't work for the VPS, it's a
+# completely separate k3s cluster with no federation to home's API server.
+# AppRole is the standard way to authenticate an external/non-native
+# client to Vault instead.
+#
+# Reachable via https://vault.kube.huskypup.net - a NEW public Cloudflare
+# A record (home was previously deliberately kept off the public
+# internet - this was an explicit, confirmed decision, not a default).
+# Goes through the same Istio ingress gateway that already serves other
+# public *.kube.huskypup.net hosts, so no new UniFi port-forward/NodePort
+# needed, and no Istio PeerAuthentication/AuthorizationPolicy change
+# either - unlike the ha-authentik-postgres NodePort case, this traffic
+# arrives already wrapped in a normal in-mesh call from the ingress
+# gateway's own identity, not raw external TCP straight to a pod.
+#
+# NOTE: the AppRole's issued token/secret_id CANNOT be IP-bound
+# (token_bound_cidrs / secret_id_bound_cidrs) over this path - confirmed
+# live 2026-08-20 that Vault only ever sees the ingress gateway's own pod
+# IP for any request arriving this way, never the VPS's real source IP.
+# Security boundary here is AppRole credential secrecy + the narrow
+# read-only secret/vps/* policy (vps-eso-reader), not network-level
+# restriction - this is the standard/expected shape of AppRole auth for
+# external clients generally, CIDR-binding is normally extra
+# defense-in-depth on top rather than the primary mechanism.
+#
+# One-time manual bootstrap on Vault's side (already done 2026-08-20, not
+# scripted - matches every other Vault policy/auth-method setup in this
+# repo, which are also applied by hand via `vault` CLI, not GitOps):
+#   vault auth enable approle
+#   vault policy write vps-eso-reader - <<'EOF'
+#   path "secret/data/vps/*"     { capabilities = ["read", "list"] }
+#   path "secret/metadata/vps/*" { capabilities = ["read", "list"] }
+#   EOF
+#   vault write auth/approle/role/vps-eso token_policies="vps-eso-reader" \
+#     token_ttl=1h token_max_ttl=4h secret_id_num_uses=0
+#   vault read auth/approle/role/vps-eso/role-id      # -> roleId below
+#   vault write -f auth/approle/role/vps-eso/secret-id  # -> secret_id
+#
+# vault-approle-creds is a plain Secret created manually on the VPS
+# (kubectl, not git - same reasoning as every other VPS secret):
+#   kubectl -n external-secrets create secret generic vault-approle-creds \
+#     --from-literal=role_id=<from role-id above> \
+#     --from-literal=secret_id=<from secret-id above>
+# Rotate the secret_id periodically by writing a new one and patching this
+# Secret - role_id is stable and not sensitive on its own (useless without
+# a valid secret_id).
+apiVersion: external-secrets.io/v1
+kind: ClusterSecretStore
+metadata:
+  name: vault-backend
+spec:
+  provider:
+    vault:
+      server: https://vault.kube.huskypup.net
+      path: secret # KV v2 mount, same as home's vault-backend
+      version: v2
+      auth:
+        appRole:
+          path: approle
+          roleId: ecaf2eda-d922-f7cf-1143-690bbbb4d8ea # not sensitive alone, see note above
+          secretRef:
+            name: vault-approle-creds
+            namespace: external-secrets
+            key: secret_id

infrastructure/vps-standby/authentik/manifests/pg-authentik-app-externalsecret.yaml

@@ -0,0 +1,29 @@
+# Multi-site active failover pilot - VPS's half of the Vault pipeline
+# (infrastructure/authentik/manifests/ha-postgres-app-pushsecret.yaml has
+# home's half, which pushes the authoritative password into Vault at
+# secret/vps/pg-authentik-app). Pulls it back down here.
+#
+# Merge (not Owner/Replace): only overwrites the `password` key, leaving
+# CNPG's own generated host/dbname/username fields on this secret intact
+# - those are correctly LOCAL to each site (this secret's `host` key
+# points at the VPS's own local -rw service, which CNPG itself still
+# needs internally, even though the app tier reads a different value via
+# the pg-authentik.ha.huskypup.net override in values.yaml).
+apiVersion: external-secrets.io/v1
+kind: ExternalSecret
+metadata:
+  name: pg-authentik-app-password
+  namespace: authentik
+spec:
+  refreshInterval: 5m
+  secretStoreRef:
+    name: vault-backend
+    kind: ClusterSecretStore
+  target:
+    name: pg-authentik-app
+    creationPolicy: Merge
+  data:
+    - secretKey: password
+      remoteRef:
+        key: vps/pg-authentik-app
+        property: password