Files
jakman85andClaude Opus 4.8 cdb546bc6e fix(cert-manager): DNS-01 self-check via public resolvers (split-horizon)
The wildcard cert stalled: challenge Presented=true (Cloudflare TXT created OK —
token works) but propagation self-check looped 'DNS record not yet propagated'.
Verified the record IS publicly resolvable (1.1.1.1/8.8.8.8); the NAT'd k3s-dev
cluster's own DNS just can't see it. Set dns01RecursiveNameservers=1.1.1.1:53,
8.8.8.8:53 + dns01RecursiveNameserversOnly=true so the self-check bypasses cluster
DNS. Realises the pre-existing TODO in this file.

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
2026-08-17 01:44:28 -05:00
..

platform/cert-manager

cert-manager — trusted TLS via Let's Encrypt, ACME DNS-01 (Cloudflare).

What this stub is

Flux HelmRelease + two ClusterIssuers (letsencrypt-staging, letsencrypt-prod — both mandated by ADR-0006). DNS-01 against the keithjackson.dev zone validates domain control without exposing the service (works for internal-only services, R-TLS-03) and enables the wildcard *.home.keithjackson.dev cert defined in ../gateway.

Contract (Issuer + Gateway TLS)

Part of the "Issuer + Gateway TLS" interface (roadmap/workstreams.md, WS-SEC + WS-NET): the wildcard Certificate (in gateway/) references the issuer named by ${ACME_CLUSTER_ISSUER} and its secret is consumed by the Gateway HTTPS listener.

Secrets

The Cloudflare API token is referenced by name (cloudflare-api-token) and committed SOPS-encrypted (WS-SEC, OQ-4, ADR-0007). No plaintext token in git.

TODO (later phases)

  • OQ-4 (resolved to Cloudflare): WS-SEC provisions the scoped Cloudflare token, SOPS-encrypted.
  • Set the ACME account email.
  • Phase 2 (WS-SEC): issue the wildcard cert; prove a trusted, non-self-signed cert at *.home.keithjackson.dev (R-TLS-01/03, Phase-2 exit gate).