Phase 0 clarification follow-ups + reconcile CI integration seams

Adopt the spec's Phase-0 clarifications (spec now at pin a7fc19f) into the impl: - schema v1.1: enforce exposure:public => auth.mode!=none (positive+negative tested). - Kyverno: add longhorn=>size and longhorn=>NOT-reacquirable invariants (R-STATE-02/03). - Ratify the homelab.keithjackson.dev/* label-key convention (now in spec docs/03). - R-TEST-06 spec-immutability gate: ci/SPEC_PIN + ci/spec-immutability.sh + gate.sh stage + Makefile target + a Gitea job (checks spec repo clean & at pin). - touched-requirements.sh: conformance tags are the authoritative source. Reconcile 3 CI seams the CI worker flagged (gate was RED, now GREEN): - create schemas/validate.sh (SCHEMA_VALIDATE_CMD entrypoint) — L0 schema gate runs. - fix l1.sh/conformance.sh to run.sh's real interface (fixed _artifacts/ paths; no --junit/--coverage) and collect artifacts. - exec bits on new scripts. make -C ci gate => PASS (l0 skips locally w/o kyverno binary; green in CI). Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
2026-08-16 00:42:43 -05:00
1 parent 1949f4e71f
commit 8c1368977e
@@ -44,8 +44,27 @@ env:
  CLUSTER_AVAILABLE: ${{ vars.CLUSTER_AVAILABLE }}

 jobs:
+  # R-TEST-06 hard gate: the specification is read-only to agents. Fail the whole run
+  # if the spec repo is modified or has drifted from ci/SPEC_PIN. Runs first; everything
+  # else depends on it. Requires the spec repo checked out beside the impl repo — wired
+  # for when Gitea hosts both. If the spec repo is unavailable the guard SKIPS (exit 3),
+  # tolerated here (surfaced, never a fake green); a real violation (exit 1) blocks.
+  spec-immutability:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4                 # the implementation repo
+      - name: Check out the read-only spec (sibling path)
+        uses: actions/checkout@v4
+        with:
+          repository: ${{ vars.SPEC_REPO_SLUG }}  # e.g. Home-Network/home-network-k3s (set in Gitea)
+          path: ../home-network-k3s
+          fetch-depth: 0
+      - name: Verify spec immutability (R-TEST-06)
+        run: make -C ci spec-immutability || [ $? -eq 3 ]
+
  lint:
    runs-on: ubuntu-latest
+    needs: [spec-immutability]
    steps:
      - uses: actions/checkout@v4
        with: { fetch-depth: 0 }          # full history so touched-requirement diff works
@@ -11,11 +11,14 @@ CI_DIR := $(dir $(abspath $(lastword $(MAKEFILE_LIST))))

 .DEFAULT_GOAL := help

-.PHONY: help lint schema l0 l1 conformance gate all touched versions tools
+.PHONY: help spec-immutability lint schema l0 l1 conformance gate all touched versions tools

 help: ## Show this help
 	@awk 'BEGIN{FS=":.*##"} /^[a-zA-Z0-9_-]+:.*##/{printf "  \033[36m%-14s\033[0m %s\n",$$1,$$2}' $(MAKEFILE_LIST)

+spec-immutability: ## R-TEST-06: fail if the spec repo is modified or drifted from ci/SPEC_PIN
+	@"$(CI_DIR)spec-immutability.sh"
+
 lint: ## yaml/lint (no cluster; live now)
 	@"$(CI_DIR)lint.sh"

@@ -0,0 +1,6 @@
+# The spec commit (home-network-k3s main) this implementation is built against.
+# R-TEST-06: agents MUST NOT modify the spec. The spec changes only via the human/
+# System Architect; each legitimate spec advance is adopted here by bumping this pin
+# in a reviewed change. CI (ci/spec-immutability.sh) fails if the spec repo is dirty
+# or its HEAD != this pin — catching any unauthorized/unadopted spec modification.
+a7fc19fad57c70e5c7b142afa142ac16e8b13091
@@ -38,20 +38,22 @@ else
  log "no touched requirements detected — running full L2/L3 subset as a safety net"
 fi

-# Contract we ASSUME of conformance/run.sh (flag for PM to reconcile):
-#   --layer l2,l3              behavioral + fault-injection layers
-#   --trigger in-cluster-job   create/await the in-cluster Chainsaw Job (Job holds
-#                              cluster creds via its scoped SA; CI uses only a
-#                              narrow trigger credential, referenced by SECRET NAME
-#                              in the workflow: CONFORMANCE_TRIGGER_KUBECONFIG)
-#   --requirements R-A,R-B     touched subset (R-TEST-03)
-#   --junit <file> --coverage <dir>   collected artifacts feed the gate
-"$CONFORMANCE_CMD" \
-  --layer l2,l3 \
-  --trigger in-cluster-job \
-  "${req_args[@]}" \
-  --junit "${JUNIT_DIR}/l2l3.junit.xml" \
-  --coverage "${COVERAGE_DIR}"
+# RECONCILED (PM 2026-08-16): run.sh writes artifacts to conformance/_artifacts/ and does
+# NOT accept --junit/--coverage. This stage is CLUSTER-GATED (only runs when
+# CLUSTER_AVAILABLE=true, Phase 1+), so its trigger path is confirmed against a real
+# cluster in Phase 1 — TODO(Phase 1): drive the in-cluster Chainsaw Job
+# (conformance/incluster/chainsaw-job.yaml) under the narrow CONFORMANCE_TRIGGER_KUBECONFIG,
+# and confirm multi-layer (l2 + l3) selection syntax against run.sh. Artifacts are then
+# COLLECTED from the fixed location below.
+"$CONFORMANCE_CMD" --layer l2,l3 "${req_args[@]}"
+
+CONF_ART="$(dirname "$CONFORMANCE_CMD")/_artifacts"
+if [ -f "${CONF_ART}/report.junit.xml" ]; then
+  cp "${CONF_ART}/report.junit.xml" "${JUNIT_DIR}/l2l3.junit.xml"
+fi
+if [ -d "${CONF_ART}/coverage" ]; then
+  cp -r "${CONF_ART}/coverage/." "${COVERAGE_DIR}/"
+fi

 ok "L2/L3 conformance passed for touched subset"
 summary "- conformance-l2-l3: **passed**"
@@ -37,6 +37,7 @@ run_stage() {
 summary "## Merge gate (R-TEST-03)"
 summary ""

+run_stage spec-immutability "${CI_DIR}/spec-immutability.sh"
 run_stage lint        "${CI_DIR}/lint.sh"
 run_stage l0          "${CI_DIR}/l0.sh"
 run_stage l1          "${CI_DIR}/l1.sh"
@@ -46,6 +47,14 @@ echo >&2
 log "==== gate decision ===="
 BLOCK=0

+# spec-immutability (R-TEST-06): a FAIL always blocks (the spec was modified/drifted).
+# A skip means the spec repo wasn't available to verify — surfaced, not green.
+case "${OUTCOME[spec-immutability]}" in
+  fail) err  "spec-immutability: FAILED -> blocks merge (R-TEST-06)"; BLOCK=1 ;;
+  skip) warn "spec-immutability: skipped (spec repo unavailable) — not verified" ;;
+  pass) ok   "spec-immutability: green" ;;
+esac
+
 # L0/L1 stages: any failure blocks. A SKIP here means a seam is unwired — in
 # Phase 0 that is tolerated (placeholders), but surfaced.
 for s in lint l0 l1; do
@@ -28,16 +28,19 @@ if [ "${#reqs[@]}" -gt 0 ]; then
  req_args=(--requirements "$(IFS=,; echo "${reqs[*]}")")
 fi

-# Contract we ASSUME of conformance/run.sh (flag for PM to reconcile):
-#   --layer l1                 select the L1 rendering layer (no cluster)
-#   --requirements R-A,R-B     filter to the touched subset
-#   --junit <file>             write JUnit XML
-#   --coverage <dir>           write the requirement-coverage report
-"$CONFORMANCE_CMD" \
-  --layer l1 \
-  "${req_args[@]}" \
-  --junit "${JUNIT_DIR}/l1.junit.xml" \
-  --coverage "${COVERAGE_DIR}"
+# RECONCILED (PM 2026-08-16): conformance/run.sh's real interface is `--layer <l1|l2|l3>`
+# [--requirements R-A,R-B]; it ALWAYS writes artifacts to a FIXED location
+# (conformance/_artifacts/) and does NOT accept --junit/--coverage. Invoke by layer +
+# touched subset, then COLLECT its artifacts into the CI artifact dirs.
+"$CONFORMANCE_CMD" --layer l1 "${req_args[@]}"
+
+CONF_ART="$(dirname "$CONFORMANCE_CMD")/_artifacts"
+if [ -f "${CONF_ART}/report.junit.xml" ]; then
+  cp "${CONF_ART}/report.junit.xml" "${JUNIT_DIR}/l1.junit.xml"
+fi
+if [ -d "${CONF_ART}/coverage" ]; then
+  cp -r "${CONF_ART}/coverage/." "${COVERAGE_DIR}/"
+fi

 ok "L1 rendering checks passed"
 summary "- l1-render: **passed**"
@@ -0,0 +1,46 @@
+#!/usr/bin/env bash
+# ci/spec-immutability.sh — R-TEST-06 gate: the specification is read-only to agents.
+#
+# The spec lives in a SEPARATE repo (home-network-k3s). Agents implement against it and
+# never modify it; only the human/System Architect changes the spec. This gate fails CI if
+# the spec repo has uncommitted edits or its HEAD has drifted from the pinned commit in
+# ci/SPEC_PIN — catching any unauthorized (or un-adopted) spec change.
+#
+# Adopting a legitimate spec advance = bump ci/SPEC_PIN to the new spec main SHA in a
+# reviewed PR. That makes every spec movement explicit and gated, per docs/14 / R-TEST-06.
+#
+# Exit codes (per ci/gate.sh contract): 0 pass · 1 fail (blocks merge) · 3 skip (spec
+# repo not available to verify — honest skip, never fake green).
+set -uo pipefail
+CI_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+# shellcheck disable=SC1091
+source "${CI_DIR}/lib.sh"
+set +e
+
+PIN="$(grep -vE '^\s*#' "${CI_DIR}/SPEC_PIN" | tr -d '[:space:]')"
+: "${SPEC_REPO:=${REPO_ROOT}/../home-network-k3s}"
+
+if [ -z "$PIN" ]; then err "ci/SPEC_PIN has no pinned SHA"; exit 1; fi
+if [ ! -d "${SPEC_REPO}/.git" ]; then
+  skip "spec repo not present at ${SPEC_REPO} (set SPEC_REPO) — cannot verify immutability"
+  exit 3
+fi
+
+dirty="$(git -C "$SPEC_REPO" status --porcelain 2>/dev/null)"
+head="$(git -C "$SPEC_REPO" rev-parse HEAD 2>/dev/null)"
+
+rc=0
+if [ -n "$dirty" ]; then
+  err "spec repo has UNCOMMITTED changes — the spec is read-only to agents (R-TEST-06):"
+  printf '%s\n' "$dirty" >&2
+  rc=1
+fi
+if [ "$head" != "$PIN" ]; then
+  err "spec HEAD (${head:0:12}) != pinned (${PIN:0:12})."
+  err "  If this is a legitimate spec advance by the System Architect, adopt it by bumping ci/SPEC_PIN."
+  err "  Otherwise the spec was modified without authorization — reject."
+  rc=1
+fi
+
+[ "$rc" -eq 0 ] && ok "spec immutable: clean and at pinned ${PIN:0:12} (R-TEST-06)"
+exit "$rc"
@@ -8,15 +8,20 @@
 # Output (stdout): newline-separated, sorted, unique R-* IDs. Nothing else goes
 # to stdout, so callers can `mapfile`/`$(...)` it directly.
 #
-# Detection = union of two signals:
-#   1. R-* tokens literally present in the diff of changed files (tests, policies,
-#      and ServiceDefinitions carry their R-IDs per docs/14 + ADR-0011).
-#   2. A coarse path->requirement map for changes that don't self-declare IDs.
+# SOURCE OF TRUTH (reconciled, PM 2026-08-16): the conformance requirement TAGS are
+# authoritative. Each Chainsaw test declares the requirements it verifies via
+# `conformance.home.k3s/<R-ID>` labels; the coverage report (conformance/_artifacts/
+# coverage/coverage.json) is generated from those tags. This script's job is only to
+# select which tagged subset a change TOUCHES, then hand those IDs to
+# `conformance.sh --requirements ...` which runs exactly the tests carrying those tags.
 #
-# ASSUMPTION (flag for PM): the authoritative change->requirement mapping is
-# really WS-TEST's coverage tooling (tests tag their own R-IDs). This script is a
-# CI-side approximation good enough to *filter* the subset; at integration it
-# should defer to / cross-check the coverage report emitted by conformance/.
+# Detection = union of:
+#   1. R-* tokens literally present in the diff of changed files — including the
+#      authoritative `conformance.home.k3s/R-*` tags on changed tests, and the R-IDs
+#      that policies/ServiceDefinitions self-declare (docs/14 + ADR-0011).
+#   2. A coarse path->requirement fallback for changes that don't self-declare IDs.
+# The coarse map (2) NEVER overrides an explicit tag; it only widens coverage for
+# untagged changes so nothing silently escapes the gate.
 set -euo pipefail
 CI_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
 # shellcheck disable=SC1091
@@ -1,11 +1,11 @@
-# Platform label/annotation convention (PROPOSED contract)
+# Platform label/annotation convention (RATIFIED)

 These label keys are consumed by the Kyverno policies in this directory. The **values**
-are fixed by the spec; the **keys** are a convention this workstream (WS-TEST) proposes
-and are a **shared contract** with WS-NET (network/egress rendering), WS-SEC (auth), and
-WS-GITOPS (ServiceDefinition → manifest rendering). Per
-[`agents/coordination-protocol.md` §2] a contract like this needs the owning roles'
-sign-off before it is locked. **Escalated as an open question in the T-0.5 report.**
+are fixed by the spec; the **keys** are the platform convention, now **ratified in the
+spec** — `homelab.keithjackson.dev/{egress-class,auth-mode,app-namespace,baseline}` are
+documented in `docs/03-networking-isolation-and-egress.md` ("Platform label-key
+convention") as the contract that network policies, Kyverno rules, and renderers key off.
+Shared contract with WS-NET (egress rendering), WS-SEC (auth), and WS-GITOPS (rendering).

 | Label key | Applied to | Values | Fixed by | Used by policy |
 |---|---|---|---|---|
@@ -17,7 +17,7 @@ kind: ValidatingPolicy
 metadata:
  name: validate-servicedefinition
  annotations:
-    homelab.keithjackson.dev/requirements: R-GITOPS-02
+    homelab.keithjackson.dev/requirements: R-GITOPS-02,R-STATE-02,R-STATE-03
 spec:
  rules:
    - name: servicedefinition-shape
@@ -63,3 +63,17 @@ spec:
            check:
              ~.storage:
                (class == 'longhorn' || class == 'bulk'): true
+          # docs/13 v1.1: longhorn (replicated, irreplaceable) volumes MUST be sized (R-STATE-02).
+          # bulk may omit size (external mount, sized outside the cluster).
+          - message: "a `longhorn` volume requires an explicit `size` (docs/13 v1.1; R-STATE-02)"
+            check:
+              ~.storage:
+                (class != 'longhorn' || size != null): true
+    - name: servicedefinition-backup
+      # docs/13 v1.1: irreplaceable (longhorn-tier) state MUST NOT be `reacquirable` — it must
+      # reference an off-cluster backup schedule (R-STATE-03). Only bulk-tier data may be reacquirable.
+      assert:
+        all:
+          - message: "backup.policy 'reacquirable' is not allowed when any volume is on the longhorn (irreplaceable) tier; reference an off-cluster schedule (docs/13 v1.1; R-STATE-03)"
+            check:
+              (backup.policy == 'reacquirable' && storage != null && contains(storage[].class, 'longhorn')): false
@@ -2,7 +2,7 @@
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "$id": "https://home.keithjackson.dev/schemas/servicedefinition.schema.json",
  "title": "ServiceDefinition",
-  "description": "Operator-facing onboarding contract for the home k3s + Longhorn platform. One ServiceDefinition per file under catalog/. Captures INTENT; the platform compiles it into concrete resources (see schemas/renderer-contract.md). Canonical source of truth: spec docs/13-service-catalog.md. Serves R-GITOPS-02; ties into R-NET-*, R-STATE-*, R-TLS-*, R-DNS-*, R-OBS-*, R-UPD-*, R-IDP-*. Schema version: 1.0.0.",
+  "description": "Operator-facing onboarding contract for the home k3s + Longhorn platform. One ServiceDefinition per file under catalog/. Captures INTENT; the platform compiles it into concrete resources (see schemas/renderer-contract.md). Canonical source of truth: spec docs/13-service-catalog.md. Serves R-GITOPS-02; ties into R-NET-*, R-STATE-*, R-TLS-*, R-DNS-*, R-OBS-*, R-UPD-*, R-IDP-*. Schema version: 1.1.0 (adds shape invariant exposure:public => auth.mode!=none, docs/13 v1.1 / docs/15 / R-IDP-02).",
  "type": "object",
  "additionalProperties": false,
  "required": ["name", "source"],
@@ -244,6 +244,20 @@
          }
        }
      }
+    },
+    {
+      "$comment": "public exposure must not be unauthenticated: an explicit auth.mode 'none' is disallowed on public services (docs/13 v1.1, docs/15, R-IDP-02). Absent auth defaults to forward-auth at render time, so it is permitted here.",
+      "if": {
+        "required": ["exposure"],
+        "properties": { "exposure": { "const": "public" } }
+      },
+      "then": {
+        "properties": {
+          "auth": {
+            "properties": { "mode": { "not": { "const": "none" } } }
+          }
+        }
+      }
    }
  ]
 }
@@ -0,0 +1,15 @@
+#!/usr/bin/env bash
+# schemas/validate.sh — validate ServiceDefinition YAML against the JSON Schema.
+# The CI L0/L1 gate's entrypoint (ci/lib.sh SCHEMA_VALIDATE_CMD). No cluster needed
+# (R-TEST-02). Usage:
+#   validate.sh <file>...    # exit 0 = all valid, non-zero = a file failed
+#   validate.sh --version    # print the validator version
+set -euo pipefail
+HERE="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+SCHEMA="${HERE}/servicedefinition.schema.json"
+
+if [ "${1:-}" = "--version" ]; then
+  exec check-jsonschema --version
+fi
+[ "$#" -gt 0 ] || { echo "usage: validate.sh <file>... | --version" >&2; exit 2; }
+exec check-jsonschema --schemafile "$SCHEMA" "$@"