Collection verbs cannot combine with resourceNames (grant would
be void). Instance verbs stay name-scoped to adguard-deployment;
list/watch is namespace-scoped (single Deployment in ns).
- grant list+watch on adguard-deployment (rollout status hung
without it, job hit activeDeadline and failed)
- compare content digests only (old hash embedded filenames, so
every run patched + restarted even when in sync)
Keeps explicit rollout restart alongside reloader annotation:
one extra restart per rotation (~60d) is accepted for
determinism if reloader is down.
adguard.forust.xyz is NXDOMAIN (host retired); the combo SAN cert kept
failing and dns.forust.xyz served TRAEFIK DEFAULT CERT. Scope prod route
to dns.forust.xyz only.
Also commit live traefik-values state (remove letsencrypt-staging
resolver, live since helm rev 33).
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)
Protect public Traefik routes with CrowdSec HTTP decisions and restore access logging for web traffic analysis.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
services marked k8s/active are applied via kubectl; the rest via docker
compose. inactive services with k8s/ keep only routing manifests
(external Services, EndpointSlices, Ingresses) to reach docker backends.
headscale/nextcloud routing moved to k8s/routing/.
validations: compose config --quiet + kubectl apply --dry-run=client.
namespace manifests applied first. pull_policy:build stacks get
build+push before up so the registry image stays fresh.