Add R-DNS-04, R-DNS-05 and R-SEC-04 to the requirement mirror and tag the
L0 tests that verify them. Align the R-ENV and R-UPD-02 summaries with the
amended requirements. Point the VPN egress docs at ADR-0020; the drafted
"ADR-0016" number went to the Cloudflare Tunnel.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
new-peer.sh generated MTU 1420, which stalled traffic on mobile data
despite a healthy handshake (one phone, one carrier, 2026-11-09). Default
to 1280, overridable through WG_MTU.
Correct stale claims in the runbook and the DaemonSet header against
live BPF conntrack: LoadBalancer VIPs see 10.90.0.x as world, not a
SNATed host identity. Add a troubleshooting section.
The DaemonSet render is unchanged, so WireGuard pods do not roll.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Add the at-home peer (10.90.0.3) and list ${WG_CLIENT_CIDR} in the PiHole
DNS frontends' loadBalancerSourceRanges. wg0 is a Cilium device, so tunnel
clients reach the source-range check untranslated; a LAN-only list is
predicted to drop their DNS. Adds an L0 guard that both ranges are present
and every range is RFC1918, and corrects the README's SNAT and IPv6 claims.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Public names are proxied CNAMEs to the Cloudflare Tunnel since PR #156 and
work from the LAN, so they should resolve publicly there too. Replace the
address=/${BASE_DOMAIN}/ wildcard with explicit entries: the LAN-only names
(derived from every controller:none HTTPRoute), gitea/nextcloud (off
Cloudflare's 100 MB / 100 s limits), keycloak/auth, and the L4 VIPs.
Add L0 check pihole-lan-names, mutation-tested. Correct docs that still
described public names as CNAME to wan. and the cloudflared entrypoint comment.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Gateway target -> fc81938b-...cfargotunnel.com, --cloudflare-proxied globally,
L4 service publishers (unifi, crafty-mc) pinned DNS-only, Plex web UI and
Headlamp disowned (LAN/VPN-only). L0 internal-services-publish-no-public-dns
gains assertion (9) tying the Gateway target to the tunnel UUID, requiring
cloudflare-proxied=false on every service-source publisher, and keeping the
tunnel's 404 list and the unpublished set consistent. L2 dns-exposure fixture
opts out of proxying.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
The pod ran as 65532:65532 with no fsGroup, so the kubelet laid the Secret's
credentials.json down root:root 0440 — unreadable to the container. cloudflared
logged "permission denied" and then the red-herring "Error locating origin cert".
- deployment.yaml: securityContext.fsGroup 65532 (mode stays 0440)
- L0 cloudflared-ingress-rules (7): Secret volumes must be readable by the
container's run-as uid/groups; mutation self-tested
- runbook: permission-denied troubleshooting entry, credential rotation section
- decisions/2026-11-08-cloudflared-creds-permission.md
Co-Authored-By: Claude Opus 5.5 <[email protected]>
servicemonitor-ports-resolve could not audit a PodMonitor with no in-repo
Service, and no-plaintext-secrets requires a template's secret value to be a
bare placeholder.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Adds platform/cloudflared: a locally-managed tunnel whose ingress rules live in
git. LAN/VPN-only hosts (plex, plex-vip, jackett, pihole, pihole-dns, semaphore,
qbittorrent, radarr, sonarr, headlamp, unifi, crafty-mc) map to 404. The apex
and *.home go to cilium-gateway-shared.gateway over TLS with matchSNItoHost.
Last comes a catch-all 404. Nothing routes to the Kubernetes API.
- 2-replica Deployment, hard node spread, non-root, pinned digest, PDB
- default-deny plus an egress CNP: Cloudflare edge 7844/443, the Gateway, and
Gateway backends by gateway-access=shared minus the LAN-only namespaces
(needed because of the CVE-2025-30162 L7LB egress re-check)
- PodMonitor plus two alerts that fire on absent series, with promtool tests
- its own Flux Kustomization (dependsOn platform-config, wait:false, nothing
depends on it), plus SOPS decryption patches in both overlays
- placeholder credentials (SOPS-encrypted) so it can merge before enrolment
- L0 cloudflared-ingress-rules, and a cloudflared_ scrape owner
- runbooks/cloudflare-tunnel.md and decisions/2026-11-08-cloudflare-tunnel.md
Co-Authored-By: Claude Opus 5.5 <[email protected]>
The 26.x Keycloak operator creates its own ServiceMonitor named after the
CR (keycloak/keycloak) when metrics are enabled; spec.serviceMonitor.enabled
defaults true. That collides with the Flux-owned platform/keycloak/
servicemonitor.yaml: the operator's PATCH 409s, it exhausts retries, the CR
stays Ready=False/HasErrors=True, and platform-config (wait: true) stalls
every downstream Kustomization.
Keep the repo's ServiceMonitor (conformance L0/L2 and the dashboard are
written against it; the operator's scrapes the same port 9000 /metrics) and
set spec.serviceMonitor.enabled: false.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Run 500 (this PR's first CI run) failed L0 prod-baremetal-not-dev-surface: act runs steps with
umask 0000, the checkout was world-writable, and Ansible (now present, via the toolbox) refuses
a world-writable directory as an ansible.cfg source. Reproduced and fixed locally in the pinned
image.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
The gate job now runs in gitea.home.keithjackson.dev/home-network/toolbox@sha256:83a3501e...
(the image images.yaml publishes from infra/devcontainer/Containerfile) instead of
catthehacker/ubuntu:act-22.04 plus a per-run download of yamllint, check-jsonschema, kyverno,
promtool and chainsaw. Local (dev.sh) and CI now gate on the identical toolchain.
- actions/checkout (Node) replaced by a plain-git step with fetch-depth-0 semantics; the
toolbox has no Node. The job token is not persisted into .git/config.
- The install step is replaced by a check that every ci/versions.env pin equals the image's
/etc/dev-toolchain-versions, so the pins stay enforced.
- Renovate follows the digest through a `# renovate-digest:` marker; github-actions manager
disabled for own images.
- install-tools.sh kept (merge-gate.yaml blueprint + alert-rules-unit-tested check use it).
Co-Authored-By: Claude Opus 5.5 <[email protected]>
The in-cluster runner ran at the default capacity 1 with no config file, so
gate / images / renovate runs on different refs queued behind each other
(29-759s waits; run 472 waited 759s behind a renovate burst).
- ConfigMap act-runner-config with ONLY runner.capacity: 2, selected via
CONFIG_FILE (honoured by the v0.6.1 image's run.sh). v0.6.1 LoadDefault()
applies the same defaults to a partial file as to no file; the
generate-config example is deliberately not used (its labels would override
the registered ones, force_pull differs from the default).
- Pod-template config-revision annotation so ConfigMap-only edits can roll it.
- dind: cpu request 100m -> 2 (no limit), memory 256Mi/2Gi -> 1Gi/4Gi.
Measured at capacity 1: peak 0.69 core, 447Mi working set (30d), 0 OOMs.
- runner: memory 128Mi/512Mi -> 256Mi/1Gi (peak 314Mi at capacity 1).
apps/wave-1/gitea-actions.yaml is a hand-owned EXTRA_UNITS file (the
renderer only places it in a wave), so it is the source, not generated output.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
- Replace the deprecated top-level `docker` key with the packageRule Renovate's
own migration produced (matchCategories docker, enabled true), keeping every
comment. Supersedes #139, which dropped all comment lines.
- Add gitea/runner (act_runner's upstream rename) to the blast-radius rule.
- rangeStrategy bump for flux/helm `>=A <B` ranges: the default `auto` resolves
to `replace`, which turned `>=1.12.0 <1.13.0` into `<1.14.0` (#131, #130,
#115). In-range `bump` updates are disabled; Flux adopts those unaided.
- Restore headlamp's lower bound: `<0.47.0` -> `>=0.46.0 <0.47.0` (live 0.46.0).
Co-Authored-By: Claude Opus 5.5 <[email protected]>
PR #121 bumped the three keycloak-k8s-resources URLs to 26.8.2, but 26.8.2
ships two more CRDs (keycloakoidcclients, keycloaksamlclients). The new
operator fails to start its keycloaksamlclients informer (Not Found) and
CrashLoops, holding Flux `platform` in health checks.
- add both CRD URLs with the same renovate marker/depName
- patch operator memory 450Mi -> limit 768Mi / request 512Mi (26.0.6 was
OOMKilled at 421/450Mi, 154 restarts)
- README/renovate comments: Renovate cannot add CRDs a release introduces
Co-Authored-By: Claude Opus 5.5 <[email protected]>
PR #136 bumps the renderer's OAUTH2_PROXY_IMAGE; the container-workload
and egress-class (vpn + internet) goldens still pinned v7.8.1. Regenerated
with the check scripts' own invocation; each diff is the image tag only.
Also updates the crafty README's current-pin row.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
The velero CLI and server are version-coupled, so VELERO_VERSION follows the
server running on prod (docker.io/velero/velero:v1.18.2, chart 12.2.3,
verified 2026-11-07). It has no renovate marker on purpose: bump it together
with the chart's appVersion. The verify stage asserts that
`velero version --client-only` reports the pin. The build arg reaches CI
through build-image.sh --print-build-args, so no .gitea/ or ci/ change is
needed.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Gitea dropped the "1." prefix; 28.x is the successor of 1.27.
- image 28.3.0, update.range >=28.0.0 <29.0.0, Renovate ceiling <29.0.0
- drop GITEA__server__DOMAIN (no longer read in 28, ROOT_URL is the source)
- declare RUN_RETENTION_DAYS=400, EGRESS_MODE=lax (no allow list: the OIDC
login source is not gated by the egress policy in 28.3.0), and
DISABLE_REGISTRATION=true (28's new default)
- document autoRollbackOnFailedHealth as inert (no ImagePolicy is rendered);
rollback is a PVC restore because the DB migration is one-way
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Add assertion (8) to internal-services-publish-no-public-dns. The external-dns
HelmRelease must declare annotationPrefix explicitly, and every
external-dns annotation in apps/, platform/, runbooks/ and catalog/render.py
must use that prefix. The v0.23.0 default-prefix change on 2026-11-07
ignored the controller: none, target and hostname annotations while
assertions (1)-(3) stayed green.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
external-dns v0.22 changed its default annotation prefix to
external-dns.kubernetes.io/ with no fallback. Renovate #129 moved the chart
to 1.23.0 (app v0.23.0), and at 12:13Z it ignored every alpha-prefixed
annotation in this repo: the Gateway target (public names went to A
192.168.2.20), controller: none (jackett, pihole and semaphore got
published), and the L4 Service hostnames (unifi and crafty-mc were deleted).
Set annotationPrefix: external-dns.alpha.kubernetes.io/ to restore the v0.21
reading. Floor the chart at 1.23.0, because the flag needs app >= 0.22.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Dispatcher image and PM_WORKER_IMAGE both move to
sha256:14a42ba8281ecd98005f5d99ecda4fb729104c783b01584097c32e104ced1bb2, the digest the
registry returns for agent-runner:sha-466f63ec3c3d and :main. config-generation 4 -> 5.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
`claude --print --output-format json` reports its own failures (API error, usage limit,
error_max_turns) as the result envelope on stdout and leaves stderr empty, so ask_pm()'s
stderr-only message logged "claude exited 1: " with nothing after it, twice on 2026-10-31.
describe_cli_failure() now logs the envelope's diagnostic fields plus stdout/stderr tails,
redacted (sk-ant-, Bearer/token, URL userinfo). PM_DECISION_MAX_TURNS (default 1, unchanged)
makes the max-turns remedy a ConfigMap change if the envelope turns out to say error_max_turns.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
The run-383 worker image (df36d9b, pinned in #96) bakes imagePullSecrets: pm-agent-registry,
which admission requires, and the dispatcher's dry-run triage of #97 was read and approved.
PM_REQUIRE_LABEL=agent and the worker caps are unchanged. The dry-run status text in the
configmap comments, the README and the plan's follow-up 3 is updated to match.
Pre-flight (read-only, k3s-prod): pm-agent-job-shape Ready+Enforce; SA/Role/RoleBinding
present and can-i create jobs = yes; pm-agent-registry and pm-agent-credentials exist; both
egress CNPs Valid.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
c4eba5b (run 340) bakes a worker template without imagePullSecrets, so the new admission rule
refuses worker Jobs. df36d9b is agent-runner:main == :sha-ddd04d460ef6 (measured on the registry),
and its /etc/pm-agent/worker-job-template.yaml carries imagePullSecrets: pm-agent-registry (read
from the published layer).
Renovate found df36d9b but holds it under "Awaiting Schedule" on dashboard #91: the operator's
dispatch (run 385) was a real run at 12:14 New York, outside the 03:00-06:00 window. The
renovate-digest markers are unchanged, so later bumps remain Renovate's.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
pihole/pihole (no tag) -> 2026.10.0 sha256:d3413471...
lscr.io/linuxserver/jackett:latest -> v0.24.2865-ls7 sha256:4fcf3cc2...
plexinc/pms-docker:latest -> 1.43.4.10903-e5521bd8c sha256:e0ab2739...
Each tag was resolved against its registry and matched the pods' imageID, so the image is
unchanged (the pod template string does change, so Flux rolls the three workloads).
renovate.json5: regex versioning for jackett (-lsN) and plex (build-hex); jackett is looked
up on ghcr.io via overridePackageName, because lscr.io is capped at dockerMaxPages=20 pages
and jackett's 32k tags put the running one at #31,980 (dry run proposed nothing without it);
jackett weekly (13 builds in 12 days); pihole deliberately unheld, year rollover = major.
Stale "publishes only latest" prose fixed in renovate.json5 and three runbooks.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Operator decision. Both are held at every update type and labelled needs-review:
- quay.io/oauth2-proxy/oauth2-proxy (render.py constant, rendered sidecars, crafty). Since v7.13.0
oauth2-proxy validates refreshed OIDC sessions against the access token, every sidecar uses
--cookie-refresh, and that path is untested against this Keycloak.
- ghcr.io/renovatebot/renovate (and renovatebot/github-action if ever reintroduced). Nothing
re-validates renovate.json5 against the new version before it takes effect.
The validator passes on 44.167.0. Evaluated with Renovate's own packageRules engine: exactly 2 of 29
branch rows change (automerge true -> false, +needs-review); the other 27 are identical.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Audit of the first completed dry runs (353/354) plus a debug-level local reproduction:
- App-image updates edited only the generated apps/wave-*. A ServiceDefinition has no apiVersion,
so the kubernetes manager extracts nothing from catalog/, and every such PR would fail
render-golden (reproduced: GOLDEN DRIFT). New regex managers for catalog source.image and
render.py OAUTH2_PROXY_IMAGE put the source and generated edits in one branch (render-golden
rc 0, reproduced).
- CNPG Cluster imageName was untracked (3 Clusters), while the grant Job was tracked. New manager.
- allowedVersions mirror catalog update.range: this drops gitea 28.x, qbittorrent 20.x, radarr 6,
nextcloud 35 and postgresql 18.
- Never auto-merge: gitea, act_runner, postgresql (blast radius), flux range widening, python
minors. Majors wait for a Dependency Dashboard tick.
- velero plugin RC: semver versioning (v1.14.3-rc.2 -> v1.14.4 now proposed).
- flux docker duplicates disabled (the "Cannot find replaceString" line). Two false comments
corrected.
Local dry run: regex deps 7 -> 20, branches 33 -> 29. The validator passes on 44.167.0.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Run 350 cleared platform init (/user 200, /version 200, /repos 200), then stopped at
GET /api/v1/orgs/Home-Network with Gitea's own 403:
token does not have at least one of required scope(s), required=[read:organization],
token scope=read:package,write:repository,read:user
The workflow's header listed only those three scopes, and the operator minted exactly that.
Renovate's initRepo -> isOrg needs read:organization, and Gitea files /issues and /labels under
the issue scope, which the dashboard and PR labels need: write:issue.
- renovate.yaml: corrects the scope list (read:user, read:organization, write:repository,
write:issue, read:package), with the call each serves. A new preflight step probes them with
GETs, the token passed to curl on stdin, and names a missing scope instead of Renovate's bare
"Authentication failure".
- ci/lint.sh: rejects workflow `uses:` lines that need a Docker daemon (docker://,
renovatebot/github-action, docker/*-action), the class fixed in #84.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
renovatebot/github-action is a wrapper around `docker run`. On this runner a job has no Docker
daemon by decision (decisions/2026-10-29-image-pipeline.md), so every run, including dispatch
run 343, died before Renovate started:
failed to connect to the docker API at unix:///var/run/docker.sock ... no such file or directory
The pinned image ghcr.io/renovatebot/renovate:44.167.0 is now the job container, and a run step
execs renovate-entrypoint.sh. actions/checkout is dropped, since Renovate clones the repo itself.
An optional RENOVATE_GITHUB_COM_TOKEN secret feeds GITHUB_COM_TOKEN.
Local dry run (--platform=local, uid 12021, the step body verbatim): 179 deps extracted, 42
updates, and all 4 own-image digest lookups against the Gitea registry succeed.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Rebased #61 onto the CI image pipeline (#81). This commit makes the remaining in-git
blockers to going live go away:
- deployment.yaml and PM_WORKER_IMAGE pin images run 340's digest (sha256:c4eba5b3...,
tag sha-523cd9118f95 = main), read from the registry. Renovate markers unchanged.
- PM_DRY_RUN=true and PM_REQUIRE_LABEL=agent for the first run. Dry run is required, not
just cautious: the run-340 image's baked worker template predates imagePullSecrets, so
admission would deny its Jobs until CI re-publishes after this merge and Renovate bumps.
- Token scopes: Gitea 1.24.3 scopes GET /api/v1/user (the dispatcher's first call) under
read:user. The documented list was missing it. Now read:user, write:repository,
write:issue, read:package everywhere.
- Org-package read documented as a team (Packages=Read, no repos), not just a scope.
- sops is toolbox-only; the .examples say so and give the in-place edit.
- runbook: key-name check no longer emits Secret values before the pipe.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
The registry had two blockers. #59 fixed the first (nodes resolved the hostname to the
non-hairpinned WAN and failed TLS against the router's CN=%D cert). This is the second:
the pull is unauthenticated, and this registry refuses unauthenticated clients outright —
measured 2026-10-05, its anonymous token carries access: null and every path 401s.
pm-agent-registry is a dockerconfigjson Secret (SOPS, placeholders) referenced by
imagePullSecrets on the dispatcher Deployment and on the committed worker Job template. It
holds the SAME Gitea token as pm-agent-credentials, because the two consumers need two
formats: the application reads a raw token from env, the kubelet reads a dockerconfigjson
before any container starts. Same one-value-in-two-files pattern as the Keycloak client
secret, documented in both .example files. The operator's part is the value plus one new
scope: read:package.
ADMISSION GAINED A SEVENTH RULE. imagePullSecrets is a pod-spec field no container
references, which is exactly why it is easy to leave unread — the same shape of omission
as initContainers, ephemeralContainers, env and envFrom before it. The kubelet reads the
named Secret, in this namespace, before anything runs. Rule 1's pattern now REQUIRES the
committed entry; rule 7 forbids any other and any second one. Both halves are needed: a
foreach over an absent list yields no result.
The leverage is small and the policy says so: the image is digest-pinned, so a different
pull credential cannot change WHAT runs. Pinned anyway — 'only the committed shape' is the
premise, and the rule costs four lines.
kyverno test: 33 passed, 0 failed (was 28). Three new fixtures, each a real path: the
committed entry plus a second one, a different Secret, and none at all.
The gate also caught me shipping the Secret in plaintext — I never ran sops -e -i on the
placeholder. Recorded in the handoff entry, because the file looked finished.
Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
Images run 337's push failed with "403 Forbidden: Access denied". Measured
read-only: Gitea never logged the request; an unauthenticated GET /v2/ via the
Gateway gets 401 from Gitea when sent from act-runner (egress unrestricted) but
403 from buildkitd; Hubble logs it as an L7 http-request DROPPED with
traffic_direction EGRESS. Cilium >= 1.16.8 (CVE-2025-30162) re-checks the
source pod's egress against the proxied backend identity, and buildkitd's
egress only allowed world and the ingress entity. Same fix as 68690e5 for
Keycloak. Gitea's ingress already admits reserved:ingress and is unchanged.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
R-TLS-01/docs/08 govern externally reachable TLS and ADR-0006 puts workload mTLS out of scope, but
tls-lifecycle-automated required ACME for every Certificate, which platform/buildkit's internal CA
cannot satisfy. Clause (5b) accepts the tls-scope=workload-mtls annotation only when the
certificate proves it: no Gateway serves it, internal names only, namespaced selfSigned/ca issuer
backed by an exempt CA certificate. Also fixes yamllint comma spacing in the buildkitd mounts.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Operator decision on PR #81: no CI job may reach a privileged daemon. The act runner is untouched.
- platform/buildkit (+ clusters/base/buildkit.yaml): moby/buildkit v0.34.1-rootless pinned by
digest, uid 1000, not privileged; caps drop ALL + SETUID/SETGID, seccomp/AppArmor Unconfined,
read-only rootfs (each measured under podman). cert-manager CA + server/client certs, ClusterIP
:1234, default-deny with ingress only from the act-runner pod (DinD masquerades jobs to it).
No Kyverno exception: every applied policy matches app-namespace=true only.
- Each Containerfile names its shipped stage and ends in a `verify` stage holding every former
`docker run` assertion (uid 65532, RUN --network=none, writable-path inventory standing in for
the read-only rootfs). The old `&& echo` checks that could never fail are real assertions now.
- images.sh drives buildkitd with a pinned buildx (sha256-checked) via bake; PRs build verify
targets with agent-runner FROM the in-run toolbox (target: context); main verifies, then pushes.
Fails loudly when the builder secrets are absent or buildkitd is unreachable.
- infra/gitea/buildkit-client-secret.sh copies the client cert into Actions secrets.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
jakman85andClaude Opus 5.5