Big migration: the source/prepared split is gone. Each zones/*.zone is
now an RFC-compliant zone file that CoreDNS reads directly. Editing a
record is just edit + bump SOA + commit. CoreDNS auto-reloads within
30s; HE pulls on its own 300s SOA-refresh cycle.
Why: groundwork for the coredns-rfc2136 plugin to edit zones in place
without juggling a source/prepared transformation step. Also reduces
the mental model from "edit source, run prep, push" to just "edit".
Changes:
- zones/*.zone: 84 files migrated from Vultr-export form to RFC-compliant
form (SOA injected, Vultr NS replaced with HE NS, CNAME/MX/NS rdata
dot-terminated, apex lines get explicit @ prefix). Diff is mechanical
and byte-count is unchanged (~340K) -- pure formatting promotion.
- docker-compose.yml: bind ./zones:/zones:ro (was ./zones-prepared)
- Makefile: dropped 'prep' target. 'reload' is now a no-op explainer.
'tls-up' no longer depends on prep. 'clean' no longer wipes prepared.
- scripts/prepare-zones.sh moved to scripts/archive/ (kept for reference).
- .gitignore: updated comment for zones-prepared/ (now legacy).
NOT in this commit (follow-ups):
- CLAUDE.md updates documenting the new workflow.
- scripts/bump-serials.sh helper for manual-edit SOA bumping.
- coredns-rfc2136 plugin refactor (Phase 2b in the plan).
The original healthcheck `wget -qO- http://127.0.0.1:8080/health` has
been failing since day one because the CoreDNS image is distroless —
no shell, no HTTP client. The container has been running in
"(unhealthy)" status the whole time without anyone noticing because
nothing depends_on it.
Replace with `/coredns -version`, which is the thinnest honest check
the image can support. For deeper liveness/readiness, scrape
:8081/health from outside the container.
Replaces the self-signed dev cert flow with a real LE prod cert for
dns.l.supported.systems, issued and auto-renewed by a Caddy sidecar
using DNS-01 challenge against the Vultr API.
Components:
- caddy/Dockerfile builds Caddy 2.10.0 with caddy-dns/vultr plugin
via xcaddy. GOTOOLCHAIN=auto so xcaddy can fetch newer Go on demand
when plugin versions advance their minimum Go.
- caddy/Caddyfile uses DNS-01 with explicit public resolvers (1.1.1.1,
9.9.9.9) for the propagation check. Without that, Docker's embedded
DNS leaks the container into the host's split-horizon LAN DNS, which
returns LAN IPs for ns1.vultr.com and the propagation check fails.
- docker-compose: caddy service shares ./caddy-data with coredns via a
read-only subpath mount that excludes /acme (account private key).
- Healthcheck doubles as a symlinker: maintains stable cert.pem /
key.pem names at /data/caddy/ and chmods cert files + their dirs to
be readable by CoreDNS's nonroot user. Flips to "healthy" only once
the symlinks dereference (i.e. cert exists), gating CoreDNS start
via depends_on: service_healthy.
- Corefile unchanged — same /etc/coredns/certs/cert.pem path; only the
bind-mount source switches from ./certs to ./caddy-data/caddy.
- New Makefile target: tls-up orchestrates the bring-up sequence.
Cert is valid until Aug 12 2026. Verified end-to-end:
dig @127.0.0.1 -p 8853 +tls +tls-hostname=dns.l.supported.systems ...
dig @127.0.0.1 -p 8443 +https +tls-hostname=dns.l.supported.systems ...
- New Corefile snippet (common) shared across plain DNS / DoT / DoH so
zone-loading + forward + cache stay DRY across all three transports
- scripts/generate-certs.sh: openssl-only self-signed RSA cert with SANs
for localhost / 127.0.0.1 / ::1 / coredns / dns.local. Idempotent —
skips regeneration if cert is valid >24h ahead; FORCE=1 to rotate.
- Key chmod is 0644 so the CoreDNS container's nonroot user can read it
via the bind mount. Acceptable for local dev; production should mount
real certs with proper UID/GID.
- DOT_PORT=8853, DOH_PORT=8443 (avoids Caddy already-on-443 collision)
- Makefile: `make certs`, `make test-tls`
- All three transports verified end-to-end (dig +tls, dig +https,
curl with raw RFC 8484 wire format)
