workstation: roster-driven provisioner (SSoT reconcile, additive-only)

t3-provision-users.sh now consumes roster_engine.py: derives accounts + per-tier groups + sticky ports + /etc/ttyd-user-map + dispatch.json from roster.yaml and applies them. ADDITIVE-ONLY for existing users (never strips a group, replaces a home, or re-locks an account) so the hourly timer is always safe. Best-effort tier validation vs live k8s_users: warns on a net-new absent user (emo), aborts only on a real tier conflict, skips when root has no Vault token. DRY_RUN mode for safe testing. Verified on the live host: reproduces dispatch.json content exactly, emo/anca groups + all t3-serve instances unchanged, idempotent, shellcheck-clean; deployed to /usr/local/bin (hourly timer target).

Engine: validate_tiers now returns ValidationIssue(severity) — error=conflict (abort) vs warn=absent (grant pending) — + has_blocking_errors(); 28 pytest cases. setup-devvm.sh redeploys the provisioner for reproducibility.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

scripts/workstation/roster_engine.py

@@ -117,26 +117,49 @@ def load_roster_file(path: str) -> Roster:
 # --------------------------------------------------------------------------
 
 
-def validate_tiers(roster: Roster, k8s_user_tiers: dict[str, str]) -> list[str]:
-    """Return one error string per roster user whose tier disagrees with the
-    live `k8s_users` map. Admins are exempt (cluster-admin is granted out of
-    band). An empty list means the roster is consistent with the cluster."""
-    errors = []
+@dataclass(frozen=True)
+class ValidationIssue:
+    os_user: str
+    severity: str  # "error" = tier conflict (abort) | "warn" = absent (grant pending)
+    message: str
+
+
+def validate_tiers(
+    roster: Roster, k8s_user_tiers: dict[str, str]
+) -> list[ValidationIssue]:
+    """Compare each roster user's tier against the live `k8s_users` map. A real
+    conflict (roster tier != cluster tier) is an "error" (abort). A net-new user
+    not yet in `k8s_users` is a "warn" (onboarding proceeds; the kubectl grant is
+    pending). Admins are exempt (cluster-admin is granted out of band). An empty
+    list means the roster is consistent with the cluster."""
+    issues = []
     for user in roster.users.values():
         if user.tier == "admin":
             continue
         actual = k8s_user_tiers.get(user.k8s_user)
         if actual is None:
-            errors.append(
-                f"{user.os_user}: tier {user.tier} but k8s_user {user.k8s_user!r} "
-                f"absent from k8s_users (add the entry first)"
+            issues.append(
+                ValidationIssue(
+                    user.os_user,
+                    "warn",
+                    f"{user.os_user}: tier {user.tier} but k8s_user {user.k8s_user!r} "
+                    f"absent from k8s_users (kubectl grant pending — add the entry)",
+                )
             )
         elif actual != user.tier:
-            errors.append(
-                f"{user.os_user}: roster tier {user.tier} != k8s_users tier "
-                f"{actual} for {user.k8s_user!r}"
+            issues.append(
+                ValidationIssue(
+                    user.os_user,
+                    "error",
+                    f"{user.os_user}: roster tier {user.tier} != k8s_users tier "
+                    f"{actual} for {user.k8s_user!r}",
+                )
             )
-    return errors
+    return issues
+
+
+def has_blocking_errors(issues: list[ValidationIssue]) -> bool:
+    return any(issue.severity == "error" for issue in issues)
 
 
 # --------------------------------------------------------------------------
@@ -261,10 +284,10 @@ def _main(argv: list[str]) -> int:
     roster = load_roster_file(args.roster)
     if args.cmd == "validate":
         with open(args.k8s_users_json, encoding="utf-8") as fh:
-            errors = validate_tiers(roster, json.load(fh))
-        for err in errors:
-            print(err, file=sys.stderr)
-        return 1 if errors else 0
+            issues = validate_tiers(roster, json.load(fh))
+        for issue in issues:
+            print(f"{issue.severity.upper()}: {issue.message}", file=sys.stderr)
+        return 1 if has_blocking_errors(issues) else 0
     with open(args.ports_json, encoding="utf-8") as fh:
         desired = derive_desired_state(roster, json.load(fh))
     json.dump(_desired_state_to_dict(desired), sys.stdout, indent=2, sort_keys=True)

scripts/workstation/setup-devvm.sh

@@ -63,4 +63,9 @@ for d in skills rules agents commands; do
 done
 log "skel: launcher + tmux + inheritance symlinks (base=$CONFIG_BASE)"
 
+# 6) deploy the roster-driven provisioner to /usr/local/bin (run hourly by
+#    t3-provision-users.timer). Re-deployed here so its logic is reproducible.
+install -m 0755 "$HERE/../t3-provision-users.sh" /usr/local/bin/t3-provision-users
+log "t3-provision-users -> /usr/local/bin/ (roster-driven)"
+
 log "OK (idempotent)"

scripts/workstation/test_roster_engine.py

@@ -99,24 +99,26 @@ def test_validate_ok_when_tiers_match():
     assert eng.validate_tiers(r, {"anca": "namespace-owner"}) == []
 
 
-def test_validate_flags_tier_mismatch():
+def test_validate_flags_tier_mismatch_as_error():
+    # roster says power-user, cluster says namespace-owner -> a real conflict -> ERROR (abort).
     r = _roster(
         "users: {ancamilea: {authentik_user: a, k8s_user: anca, tier: power-user}}"
     )
-    errs = eng.validate_tiers(r, {"anca": "namespace-owner"})
-    assert len(errs) == 1
-    assert (
-        "anca" in errs[0] and "power-user" in errs[0] and "namespace-owner" in errs[0]
-    )
+    issues = eng.validate_tiers(r, {"anca": "namespace-owner"})
+    assert len(issues) == 1
+    assert issues[0].severity == "error"
+    assert issues[0].os_user == "ancamilea"
+    assert "power-user" in issues[0].message and "namespace-owner" in issues[0].message
 
 
-def test_validate_flags_netnew_user_absent_from_k8s_users():
-    # emo is power-user in the roster but has no k8s_users entry yet -> the OIDC
-    # RBAC binding can't exist, so this must fail loud (add the entry first).
+def test_validate_flags_netnew_absent_as_warn():
+    # emo is power-user in the roster but has no k8s_users entry yet. Onboarding the
+    # workstation should still proceed; the kubectl grant is pending -> WARN, not error.
     r = _roster("users: {emo: {authentik_user: e, k8s_user: emo, tier: power-user}}")
-    errs = eng.validate_tiers(r, {})
-    assert len(errs) == 1
-    assert "emo" in errs[0] and "k8s_users" in errs[0]
+    issues = eng.validate_tiers(r, {})
+    assert len(issues) == 1
+    assert issues[0].severity == "warn"
+    assert "emo" in issues[0].message and "k8s_users" in issues[0].message
 
 
 def test_validate_skips_admin_tier():
@@ -127,6 +129,22 @@ def test_validate_skips_admin_tier():
     assert eng.validate_tiers(r, {}) == []
 
 
+def test_has_blocking_errors_distinguishes_mismatch_from_absent():
+    mismatch = _roster(
+        "users: {ancamilea: {authentik_user: a, k8s_user: anca, tier: power-user}}"
+    )
+    absent = _roster(
+        "users: {emo: {authentik_user: e, k8s_user: emo, tier: power-user}}"
+    )
+    assert (
+        eng.has_blocking_errors(
+            eng.validate_tiers(mismatch, {"anca": "namespace-owner"})
+        )
+        is True
+    )
+    assert eng.has_blocking_errors(eng.validate_tiers(absent, {})) is False
+
+
 # --------------------------------------------------------------------------
 # derive_desired_state: accounts, sticky ports, ttyd map, dispatch (module #1)
 # --------------------------------------------------------------------------