feat(adguard): sync Traefik prod cert for dns.forust.xyz into adguard-certs

CronJob adguard-cert-sync (daily 03:17) copies the public cert/key for dns.forust.xyz from Traefik acme.json into Secret adguard-certs, which AdGuard mounts for DNS-over-TLS on :853. - least-privilege RBAC: read pods/exec in ns traefik, get/update/patch Secret adguard-certs and get/patch adguard-deployment in ns adguard - script selects the PROD resolver entry only, matches main domain or SANs, compares sha256 hashes, patches the secret and restarts the deployment ONLY on change; exits non-zero and touches nothing when Traefik holds no cert yet (HTTP-01 currently cannot complete)
2026-09-23 13:39:38 +02:00
1 parent c139d700f1
commit 1e8479b853
@@ -0,0 +1,133 @@
+# AdGuard TLS cert sync: copy Traefik's public certificate for dns.forust.xyz
+# (used by DNS-over-TLS on :853) from Traefik's acme.json into Secret
+# `adguard-certs`, restarting the AdGuard Deployment only when it changed.
+#
+# Why this exists: cert-manager Certificate objects cannot be used here.
+# Traefik's acme-http@internal router hijacks every HTTP-01 challenge path,
+# so the prod ClusterIssuer can never complete an order for this host.
+# Traefik itself keeps renewing the cert via its own ACME stack; this job
+# mirrors the resulting public cert/key into the secret AdGuard mounts.
+#
+# Safety properties (all enforced by the script, not by convention):
+#   * selects the PROD resolver entry only (`.letsencrypt`), never staging;
+#   * matches by main domain OR SAN list (Traefik stores the bundled cert
+#     under the router's first domain, e.g. adguard.forust.xyz);
+#   * compares sha256 hashes and patches the Secret ONLY on change;
+#   * restarts the Deployment ONLY when the Secret was patched;
+#   * exits non-zero and touches nothing when Traefik has no cert yet,
+#     when the Secret is missing, or when the payload fails PEM checks.
+#
+# Manual apply:
+#   kubectl apply -f adguardhome/k8s/cert-sync-rbac.yaml
+#   kubectl apply -f adguardhome/k8s/cert-sync.yaml
+# Force a run (safe: idempotent, read-only when already in sync):
+#   kubectl create job -n adguard --from=cronjob/adguard-cert-sync sync-now
+apiVersion: batch/v1
+kind: CronJob
+metadata:
+  name: adguard-cert-sync
+  namespace: adguard
+  labels:
+    app: adguard
+spec:
+  schedule: "17 3 * * *"
+  concurrencyPolicy: Forbid
+  successfulJobsHistoryLimit: 2
+  failedJobsHistoryLimit: 3
+  jobTemplate:
+    spec:
+      activeDeadlineSeconds: 300
+      template:
+        metadata:
+          labels:
+            app: adguard
+        spec:
+          serviceAccountName: adguard-cert-sync
+          restartPolicy: OnFailure
+          containers:
+            - name: cert-sync
+              image: dtzar/helm-kubectl:3.19.1
+              imagePullPolicy: IfNotPresent
+              resources:
+                requests:
+                  cpu: "50m"
+                  memory: "64Mi"
+                limits:
+                  cpu: "200m"
+                  memory: "256Mi"
+              env:
+                - name: SYNC_DOMAIN
+                  value: "dns.forust.xyz"
+                - name: SYNC_RESOLVER
+                  value: "letsencrypt"
+              command:
+                - /bin/sh
+                - -c
+                - |
+                  set -eu
+                  DOMAIN="${SYNC_DOMAIN:?}"
+                  RESOLVER="${SYNC_RESOLVER:?}"
+                  NS="adguard"
+                  SECRET="adguard-certs"
+                  DEPLOY="adguard-deployment"
+
+                  echo "== 1. locate running traefik pod =="
+                  POD="$(kubectl get pods -n traefik -l app.kubernetes.io/name=traefik \
+                    --field-selector=status.phase=Running -o jsonpath='{.items[0].metadata.name}')"
+                  if [ -z "${POD:-}" ]; then
+                    echo "ERROR: no running traefik pod found, leaving secret untouched"
+                    exit 1
+                  fi
+                  echo "traefik pod: $POD"
+
+                  echo "== 2. fetch ${DOMAIN} cert/key from acme.json (resolver ${RESOLVER}) =="
+                  TMP="$(mktemp -d)"
+                  trap 'rm -rf "$TMP"' EXIT INT TERM
+                  kubectl exec -n traefik "$POD" -- cat /data/letsencrypt/acme.json > "$TMP/acme.json"
+                  jq -r --arg r "$RESOLVER" --arg d "$DOMAIN" \
+                    '.[$r].Certificates // [] | map(select(.domain.main == $d or ((.domain.sans // []) | index($d))))' \
+                    "$TMP/acme.json" \
+                    | jq -r '.[0] // empty | .certificate // empty' > "$TMP/new.crt.b64"
+                  jq -r --arg r "$RESOLVER" --arg d "$DOMAIN" \
+                    '.[$r].Certificates // [] | map(select(.domain.main == $d or ((.domain.sans // []) | index($d))))' \
+                    "$TMP/acme.json" \
+                    | jq -r '.[0] // empty | .key // empty' > "$TMP/new.key.b64"
+                  if [ ! -s "$TMP/new.crt.b64" ] || [ ! -s "$TMP/new.key.b64" ]; then
+                    echo "ERROR: no certificate for ${DOMAIN} under resolver ${RESOLVER} in acme.json."
+                    echo "HINT: Traefik has not issued it (check HTTP-01 reachability and DNS records)."
+                    echo "Leaving secret ${SECRET} untouched."
+                    exit 1
+                  fi
+                  base64 -d "$TMP/new.crt.b64" > "$TMP/new.crt"
+                  base64 -d "$TMP/new.key.b64" > "$TMP/new.key"
+                  grep -q "BEGIN CERTIFICATE" "$TMP/new.crt" || { echo "ERROR: payload is not a PEM certificate"; exit 1; }
+                  grep -q "BEGIN .*PRIVATE KEY" "$TMP/new.key" || { echo "ERROR: payload is not a PEM private key"; exit 1; }
+                  echo "fetched PEM cert/key for ${DOMAIN} (sanity checks passed)"
+
+                  echo "== 3. compare with live secret ${SECRET} =="
+                  if ! kubectl -n "$NS" get secret "$SECRET" >/dev/null 2>&1; then
+                    echo "ERROR: secret $NS/${SECRET} does not exist, refusing to create it implicitly."
+                    echo "HINT: bootstrap it once, then re-run this job."
+                    exit 1
+                  fi
+                  kubectl -n "$NS" get secret "$SECRET" -o jsonpath='{.data.tls\.crt}' \
+                    | base64 -d > "$TMP/live.crt"
+                  kubectl -n "$NS" get secret "$SECRET" -o jsonpath='{.data.tls\.key}' \
+                    | base64 -d > "$TMP/live.key"
+                  NEW_HASH="$(sha256sum "$TMP/new.crt" "$TMP/new.key" | sha256sum | cut -d' ' -f1)"
+                  LIVE_HASH="$(sha256sum "$TMP/live.crt" "$TMP/live.key" | sha256sum | cut -d' ' -f1)"
+                  if [ "$NEW_HASH" = "$LIVE_HASH" ]; then
+                    echo "secret ${SECRET} already holds the current ${DOMAIN} cert, nothing to do"
+                    exit 0
+                  fi
+                  echo "cert differs, patching secret ${SECRET}"
+
+                  echo "== 4. update secret and restart ${DEPLOY} =="
+                  CRT_B64="$(base64 "$TMP/new.crt" | tr -d '\n')"
+                  KEY_B64="$(base64 "$TMP/new.key" | tr -d '\n')"
+                  kubectl -n "$NS" patch secret "$SECRET" --type=merge \
+                    -p '{"data":{"tls.crt":"'"$CRT_B64"'","tls.key":"'"$KEY_B64"'"}}'
+                  echo "secret patched, restarting deployment"
+                  kubectl -n "$NS" rollout restart "deploy/${DEPLOY}"
+                  kubectl -n "$NS" rollout status "deploy/${DEPLOY}" --timeout=180s
+                  echo "sync complete"