feat(cicd): Slice A — pre-deploy gate + role-split migration smoke

Adds .gitea/workflows/test.yaml that gates every push on `go build`,
`bun run build`, `go vet`, the migration coordination check, and the
role-split end-to-end migration smoke. On push to main + green, calls
Dokploy's compose.deploy API and polls /health/ready until 200.

t-paliad-282 / m/paliad#114. Design: docs/design-cicd-pre-deploy-gate-2026-05-25.md
(inventor shift on mai/cronus/inventor-ci-cd-pre).

Catches all three of today's outage classes:

  brunel (~13:20) slot collision     -> TestMigrations_NoDuplicateSlot
  hermes (~16:05) dropped-col refs   -> TestBootSmoke
  mig 129 (~14:56) 42501 ownership   -> TestMigrations_EndToEndAsAppRole

Snapshot approach. internal/db/testdata/prod-snapshot.sql is a pg_dump
of youpc-supabase paliad schema + applied_migrations rows. CI restores
this into a fresh `supabase/postgres:15.8.1.060` (same image, same role
topology as prod) and runs ApplyMigrations as the `postgres` role
(which is NOT a superuser on supabase/postgres, matching prod). Existing
migrations are skipped (already in applied_migrations); only NEW migs
from the PR run end-to-end. This sidesteps the fresh-DB idempotence
debt in some historical migrations (mig 037 missing pg_trgm, mig 051
inner COMMIT) — those are tracked separately and don't block the gate.

Sub-changes:

- internal/handlers/handlers.go — new /health/ready endpoint distinct
  from /healthz. /healthz stays liveness (process alive, no DB); /ready
  is readiness (DB pool pings within 2 s). Returns 503 when svc or pool
  is nil (DB-less deploys are intentionally not-ready). svc.Pool added
  to handlers.Services, wired in cmd/server/main.go.

- internal/db/migrate_test.go — TestMigrations_NoDuplicateSlot (pure
  unit, catches brunel) and TestMigrations_EndToEndAsAppRole (snapshot-
  gated, catches the 42501 class).

- cmd/server/main_smoke_test.go — TestBootSmoke now also asserts
  /health/ready returns 503 with a nil svc. New TestHealthReady_Live
  asserts 200 against a live pool.

- internal/db/migrations/024_rename_department_columns.up.sql and
  027_rename_to_partner_units.up.sql — ALTER INDEX / ALTER POLICY
  exception handlers now catch undefined_object OR undefined_table OR
  duplicate_object. Old handler only caught undefined_object; Postgres
  raises undefined_table when source object never existed, and
  duplicate_object when destination already exists. The expanded
  handlers make these migrations truly idempotent across all plausible
  starting states.

- Makefile — verify-mig-app, test-frontend, refresh-snapshot targets.
  refresh-snapshot pg_dumps youpc-supabase prod (needs PALIAD_PROD_DATABASE_URL),
  strips pg16 \restrict commands for pg15 restore compat, and filters
  applied_migrations rows to this branch's max on-disk version.

- internal/db/testdata/README.md — explains the snapshot's purpose,
  refresh procedure, and how to verify locally.

- docs/cicd-runner-setup-2026-05-25.md — one-time admin steps for
  registering a Gitea Actions runner on mriver and wiring DOKPLOY_TOKEN
  as a repo secret. Documents soft-launch plan per m's Q11.4 (keep
  Dokploy's autoDeploy=true webhook alive for one week, disable after
  the workflow has gated 5 successful deploys).

Build clean. Full go test ./internal/... ./cmd/... green without
TEST_DATABASE_URL. With TEST_DATABASE_URL + TEST_APP_DATABASE_URL set
to a supabase/postgres scratch + snapshot restored:
TestMigrations_NoDuplicateSlot, TestMigrations_EndToEndAsAppRole,
TestBootSmoke, TestHealthReady_Live all pass. Live-DB service tests in
internal/services/* fail under supabase/postgres 15.8 with a 42P08
parameter-binding error (unrelated to Slice A — tracked as a follow-up).

.gitea/workflows/test.yaml

@@ -0,0 +1,242 @@
+# Paliad CI gate (t-paliad-282 / m/paliad#114).
+#
+# Single workflow, two purposes:
+#
+#   - On every push: gate tier — build + unit + migration smoke. Red gate
+#     means no further work and (on main) no deploy.
+#   - On push to main with gate green: deploy step — calls the Dokploy
+#     compose-deploy API for paliad's compose Zx147ycurfYagKRl_Zzyo, then
+#     polls /health/ready until the new container reports 200.
+#
+# The deploy step REPLACES the previous Gitea-push → Dokploy webhook path
+# (per m's Q11.4 pick: soft-launch with both alive for ~1 week, then
+# disable the Dokploy auto-deploy toggle). Soft-launch leaves Dokploy's
+# autoDeploy=true intact today — the workflow's deploy step is additive
+# and idempotent (Dokploy's deploy is itself idempotent).
+#
+# Catches the three failure classes from 2026-05-25:
+#
+#   - brunel slot collision (~13:20) — TestMigrations_NoDuplicateSlot,
+#     pure unit, no DB needed.
+#   - hermes dropped-col refs (~16:05) — TestBootSmoke, applies all NEW
+#     migrations (those not in the snapshot) end-to-end against a
+#     scratch DB restored from internal/db/testdata/prod-snapshot.sql.
+#   - mig 129 42501 ownership (~14:56→) — TestMigrations_EndToEndAsAppRole,
+#     applies new migrations as the prod-shaped `postgres` role (which
+#     is NOT a superuser on supabase/postgres — same shape as
+#     youpc-supabase prod, see internal/db/testdata/README.md).
+#
+# Snapshot approach: dump paliad schema + applied_migrations rows from
+# prod, commit them. CI restores → ApplyMigrations sees existing migs as
+# applied, only runs NEW migs (the ones this PR adds). This sidesteps the
+# fresh-DB idempotence requirement on historical migrations (some of
+# which use raw COMMIT or pre-installed extensions and can't be replayed
+# from scratch). To refresh: `make refresh-snapshot`.
+#
+# Design: docs/design-cicd-pre-deploy-gate-2026-05-25.md (cronus inventor
+# shift, t-paliad-282).
+
+name: Paliad CI gate
+
+on:
+  push:
+    branches:
+      - main
+      - 'mai/**'
+  pull_request:
+    branches: [main]
+
+env:
+  GO_VERSION: '1.24'
+  BUN_VERSION: '1.2'
+
+jobs:
+  # Gate job 1 — pure build. Catches go/bun build breakage that local
+  # `go build` would catch but which a worker might have skipped before
+  # pushing. Fast (~60 s) so a red here surfaces immediately.
+  build:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Set up Go
+        uses: actions/setup-go@v5
+        with:
+          go-version: ${{ env.GO_VERSION }}
+          cache: true
+
+      - name: go build
+        run: go build ./...
+
+      - name: go vet
+        run: go vet ./...
+
+      - name: Set up Bun
+        uses: oven-sh/setup-bun@v2
+        with:
+          bun-version: ${{ env.BUN_VERSION }}
+
+      - name: bun install + build
+        working-directory: frontend
+        run: |
+          bun install --frozen-lockfile
+          bun run build
+
+  # Gate job 2 — Go test suite + migration smoke against snapshot-restored
+  # scratch DB.
+  #
+  # The Postgres service container uses the same supabase/postgres image
+  # as youpc-supabase prod. The CI scratch DB starts empty; a setup step
+  # installs pg_trgm + restores the snapshot. After restore, paliad
+  # schema is at HEAD-of-snapshot and applied_migrations covers every
+  # migration up to (and including) the snapshot's max version.
+  #
+  # ApplyMigrations called in TestBootSmoke / TestMigrations_EndToEndAsAppRole
+  # sees the snapshot's applied set, finds whatever NEW migrations this
+  # PR added on top, and applies only those. The role-split smoke runs as
+  # `postgres` (which is NOT a superuser on supabase/postgres, matching
+  # the prod role topology) — any new migration that needs supabase_admin
+  # privilege fails here as it would in prod.
+  test-go:
+    runs-on: ubuntu-latest
+
+    services:
+      # supabase/postgres baked-in auth schema + supabase role topology
+      # matches youpc-supabase prod. `postgres` here is NOT a superuser
+      # (verified live: \du postgres shows "Create role, Create DB,
+      # Replication, Bypass RLS" — no Superuser). This is the prod-shaped
+      # role the deploy uses.
+      postgres:
+        image: supabase/postgres:15.8.1.060
+        env:
+          POSTGRES_PASSWORD: ci
+          POSTGRES_DB: paliad_scratch
+        ports:
+          - 5432:5432
+        options: >-
+          --health-cmd "pg_isready -U postgres"
+          --health-interval 5s
+          --health-timeout 5s
+          --health-retries 30
+
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Set up Go
+        uses: actions/setup-go@v5
+        with:
+          go-version: ${{ env.GO_VERSION }}
+          cache: true
+
+      - name: Install postgresql-client
+        run: |
+          apt-get update -qq && apt-get install -y -qq postgresql-client
+
+      # Snapshot restore. Two prep steps as supabase_admin (the actual
+      # superuser): GRANT CREATE so the `postgres` role can later create
+      # schemas if a new mig needs it; install pg_trgm so the snapshot's
+      # trigram indexes restore. Snapshot itself loads as `postgres`.
+      - name: Provision + restore snapshot
+        env:
+          PGPASSWORD: ci
+        run: |
+          set -euo pipefail
+          psql -h localhost -U supabase_admin -d paliad_scratch -v ON_ERROR_STOP=1 \
+            -c "GRANT CREATE ON DATABASE paliad_scratch TO postgres;" \
+            -c "CREATE EXTENSION IF NOT EXISTS pg_trgm;"
+          psql -h localhost -U postgres -d paliad_scratch -v ON_ERROR_STOP=1 \
+            -f internal/db/testdata/prod-snapshot.sql
+
+      # Pre-flight: catches brunel slot collision in seconds, no DB
+      # contact (still useful even though the test-go job has Postgres
+      # running, because the failure mode is independent).
+      - name: Migration coordination check
+        run: go test -count=1 -run TestMigrations_NoDuplicateSlot ./internal/db/
+
+      # Role-split end-to-end apply. Connects as `postgres` (NOT a
+      # superuser on supabase/postgres) and runs ApplyMigrations against
+      # the snapshot-restored DB. Existing migs are skipped (already in
+      # applied_migrations); NEW migs in this PR apply here. If a new
+      # migration assumes supabase_admin privilege, fails with the same
+      # 42501 error class that took paliad.de offline on 2026-05-25.
+      - name: Migration end-to-end (deploy role)
+        env:
+          TEST_APP_DATABASE_URL: postgres://postgres:ci@localhost:5432/paliad_scratch?sslmode=disable
+        run: go test -count=1 -run TestMigrations_EndToEndAsAppRole ./internal/db/
+
+      # Boot smoke. Confirms ApplyMigrations succeeds + applied set
+      # matches on-disk set + /healthz returns 200 + /health/ready
+      # returns 200 (the live-pool variant via TestHealthReady_Live).
+      - name: Boot smoke + readiness
+        env:
+          TEST_DATABASE_URL: postgres://postgres:ci@localhost:5432/paliad_scratch?sslmode=disable
+        run: go test -count=1 -run 'TestBootSmoke|TestHealthReady_Live' ./cmd/server/
+
+      # Full Go test suite WITHOUT TEST_DATABASE_URL so live-DB service
+      # tests skip (same shape as a developer laptop without a scratch
+      # DB). Live-DB tests in internal/services/* will be activated by a
+      # follow-up shift once the snapshot is verified stable across
+      # multiple PRs — they need investigation against supabase/postgres
+      # 15.8 (parameter type inference differs subtly from youpc-supabase).
+      - name: go test ./... (pure + skip-on-no-DB)
+        run: go test -count=1 ./internal/... ./cmd/...
+
+  # Deploy step. Only runs on push to main and only after both gate jobs
+  # are green. Calls Dokploy's compose.deploy with the paliad compose ID
+  # (Zx147ycurfYagKRl_Zzyo) and polls /health/ready until it returns 200
+  # or times out.
+  #
+  # Skipped on PR / feature branch pushes — those run the gate tier as
+  # a status check but don't trigger a prod deploy. Dokploy's existing
+  # autoDeploy=true webhook continues to fire during the soft-launch
+  # window (per Q11.4); it can be disabled in the Dokploy UI once this
+  # workflow has gated ≥5 successful green deploys.
+  deploy:
+    runs-on: ubuntu-latest
+    needs: [build, test-go]
+    if: github.ref == 'refs/heads/main' && github.event_name == 'push'
+
+    steps:
+      - name: Trigger Dokploy compose deploy
+        env:
+          DOKPLOY_KEY: ${{ secrets.DOKPLOY_TOKEN }}
+          DOKPLOY_API: http://100.99.98.201:3000/api/trpc
+          COMPOSE_ID: Zx147ycurfYagKRl_Zzyo
+        run: |
+          set -euo pipefail
+          if [ -z "${DOKPLOY_KEY:-}" ]; then
+            echo "ERROR: DOKPLOY_TOKEN secret is not configured."
+            echo "       Set the secret in Gitea repo settings before this step can deploy."
+            exit 2
+          fi
+          echo "==> POST compose.deploy"
+          curl -sS --connect-timeout 5 --max-time 30 \
+            -X POST \
+            -H "x-api-key: $DOKPLOY_KEY" \
+            -H "Content-Type: application/json" \
+            -d "{\"json\":{\"composeId\":\"$COMPOSE_ID\"}}" \
+            "$DOKPLOY_API/compose.deploy"
+          echo
+
+      - name: Wait for /health/ready
+        run: |
+          set -euo pipefail
+          echo "==> polling https://paliad.de/health/ready"
+          # Up to 5 minutes (60 × 5 s) — paliad's cold-start is normally
+          # ≤30 s; the longer budget covers slow image pulls + migration
+          # apply.
+          for i in $(seq 1 60); do
+            status=$(curl -sS --connect-timeout 3 --max-time 5 \
+              -o /dev/null -w '%{http_code}' \
+              https://paliad.de/health/ready || echo "000")
+            if [ "$status" = "200" ]; then
+              echo "ready after ${i} poll(s)"
+              exit 0
+            fi
+            echo "  [$i/60] status=$status — sleeping 5s"
+            sleep 5
+          done
+          echo "ERROR: /health/ready did not return 200 within 5 minutes."
+          echo "       The deploy fired but the new container is not serving."
+          echo "       Investigate: ssh mlake 'docker logs --tail 50 compose-transmit-multi-byte-driver-v7jth9-web-1'"
+          exit 1

Makefile

@@ -21,18 +21,24 @@
 # the test runner's working dirs. None of them touch internal/db/migrations/
 # files.
 
-.PHONY: help verify-migrations verify-mig test test-go
+.PHONY: help verify-migrations verify-mig verify-mig-app test test-go test-frontend refresh-snapshot
 
 help:
 	@echo "Paliad — developer targets"
 	@echo ""
 	@echo "  verify-migrations   Dry-run pending migrations + boot smoke (needs TEST_DATABASE_URL)"
 	@echo "  verify-mig          Alias for verify-migrations"
+	@echo "  verify-mig-app      End-to-end migration smoke as non-superuser role"
+	@echo "                      (needs TEST_APP_DATABASE_URL — t-paliad-282 / m/paliad#114)"
 	@echo "  test                Short test pass — covers gate tier"
 	@echo "  test-go             Full Go suite with race detector"
+	@echo "  test-frontend       Frontend bun:test suite"
 	@echo ""
 	@echo "Set TEST_DATABASE_URL to enable live-DB tests. Example:"
 	@echo "  export TEST_DATABASE_URL=postgres://paliad:...@localhost:11833/paliad_test"
+	@echo ""
+	@echo "Set TEST_APP_DATABASE_URL to enable the role-split smoke. Example:"
+	@echo "  export TEST_APP_DATABASE_URL=postgres://paliad_app:...@localhost:5432/paliad_scratch"
 
 # Gate target — the test that would have caught mig 098 / mig 099 before
 # deploy. Combines:
@@ -71,3 +77,67 @@ test:
 # (full suite, not per-PR).
 test-go:
 	go test -race ./...
+
+# Frontend bun:test suite. Runs the 4 existing pure-TS tests today; will
+# grow as mendel's Slice 3 (frontend test infill) lands.
+test-frontend:
+	cd frontend && bun test
+
+# Role-split end-to-end migration smoke — the catch for the mig 129 42501
+# ownership class (m/paliad#114). Runs ApplyMigrations as a non-superuser
+# role against TEST_APP_DATABASE_URL. Fails the build if any migration
+# assumes more privilege than the deploy role has.
+#
+# Developer setup (local):
+#   psql -c "CREATE ROLE paliad_app LOGIN PASSWORD 'ci' NOSUPERUSER;"
+#   psql -c "CREATE DATABASE paliad_scratch OWNER paliad_app;"
+#   export TEST_APP_DATABASE_URL=postgres://paliad_app:ci@localhost:5432/paliad_scratch
+verify-mig-app:
+	@if [ -z "$$TEST_APP_DATABASE_URL" ]; then \
+		echo "ERROR: TEST_APP_DATABASE_URL is not set."; \
+		echo "       The role-split migration smoke cannot run without a non-superuser scratch DB."; \
+		echo "       See Makefile comments above this target for setup."; \
+		exit 2; \
+	fi
+	go test -count=1 -run TestMigrations_EndToEndAsAppRole ./internal/db/
+
+# Refresh the prod schema snapshot used by CI's migration smoke
+# (t-paliad-282 / m/paliad#114). Connects to youpc-supabase prod, dumps
+# the paliad schema + applied_migrations rows, strips rows beyond the
+# current branch's max on-disk version, and writes
+# internal/db/testdata/prod-snapshot.sql.
+#
+# When to refresh:
+#   - After merging a PR that added a new migration to main.
+#   - When CI's migration smoke starts spuriously failing because the
+#     snapshot's applied set diverges from on-disk by more than this
+#     branch's worth of new migs.
+#
+# Requires PALIAD_PROD_DATABASE_URL env var (a Postgres URL with
+# pg_dump rights on youpc-supabase). Example:
+#   export PALIAD_PROD_DATABASE_URL='postgres://postgres:PW@100.99.98.201:11833/postgres'
+refresh-snapshot:
+	@if [ -z "$$PALIAD_PROD_DATABASE_URL" ]; then \
+		echo "ERROR: PALIAD_PROD_DATABASE_URL is not set."; \
+		echo "       Refresh requires read access to youpc-supabase prod."; \
+		exit 2; \
+	fi
+	@echo "==> dumping paliad schema (no owner, no privs)..."
+	@pg_dump --schema-only --schema=paliad --no-owner --no-privileges \
+		--no-publications --no-subscriptions \
+		"$$PALIAD_PROD_DATABASE_URL" > internal/db/testdata/prod-snapshot.sql.tmp
+	@echo "==> appending applied_migrations rows..."
+	@pg_dump --data-only --table=paliad.applied_migrations \
+		--no-owner --no-privileges \
+		"$$PALIAD_PROD_DATABASE_URL" >> internal/db/testdata/prod-snapshot.sql.tmp
+	@echo "==> stripping pg16 \\restrict / \\unrestrict commands for pg15 compat..."
+	@sed -i.bak '/^\\restrict /d; /^\\unrestrict /d' internal/db/testdata/prod-snapshot.sql.tmp
+	@rm -f internal/db/testdata/prod-snapshot.sql.tmp.bak
+	@echo "==> stripping applied_migrations rows beyond branch's max on-disk version..."
+	@MAX_VER=$$(ls internal/db/migrations/*.up.sql | xargs -I{} basename {} | sed 's/_.*//' | sort -n | tail -1); \
+	awk -v max=$$MAX_VER ' \
+	  /^[0-9]+\t/ { split($$0, a, "\t"); if (a[1]+0 > max) next; } \
+	  { print } \
+	' internal/db/testdata/prod-snapshot.sql.tmp > internal/db/testdata/prod-snapshot.sql
+	@rm internal/db/testdata/prod-snapshot.sql.tmp
+	@wc -l internal/db/testdata/prod-snapshot.sql

cmd/server/main.go

@@ -165,6 +165,7 @@ func main() {
 		sysAuditSvc := services.NewSystemAuditLogService(pool)
 		checklistTemplateSvc := services.NewChecklistTemplateService(pool, checklistCatalogSvc, sysAuditSvc, users)
 		svcBundle = &handlers.Services{
+			Pool:           pool,
 			Project:        projectSvc,
 			Team:           teamSvc,
 			PartnerUnit:    partnerUnitSvc,

cmd/server/main_smoke_test.go

@@ -98,6 +98,51 @@ func TestBootSmoke(t *testing.T) {
 	if body := strings.TrimSpace(rec.Body.String()); body != "ok" {
 		t.Errorf("GET /healthz: body=%q; want \"ok\"", body)
 	}
+
+	// (4) Readiness probe. With a nil Services bundle the endpoint MUST
+	// report 503 — that's the contract documented in handlers/handlers.go.
+	// A separate svc-with-Pool case is exercised in TestHealthReady (live).
+	rec = httptest.NewRecorder()
+	req = httptest.NewRequest(http.MethodGet, "/health/ready", nil)
+	mux.ServeHTTP(rec, req)
+	if rec.Code != http.StatusServiceUnavailable {
+		t.Errorf("GET /health/ready (nil svc): status=%d; want 503", rec.Code)
+	}
+}
+
+// TestHealthReady_Live asserts the readiness probe answers 200 when the
+// pool is reachable, 503 when it isn't. Requires TEST_DATABASE_URL.
+//
+// Why a separate test: TestBootSmoke runs Register with svc=nil to keep
+// its setup minimal; the pool-reachable path needs the pool wired in
+// through svc.Pool. Two tests, two assertions, no entanglement.
+func TestHealthReady_Live(t *testing.T) {
+	url := os.Getenv("TEST_DATABASE_URL")
+	if url == "" {
+		t.Skip("TEST_DATABASE_URL not set — skipping live readiness probe")
+	}
+
+	if err := db.ApplyMigrations(url); err != nil {
+		t.Fatalf("db.ApplyMigrations: %v", err)
+	}
+	pool, err := db.OpenPool(url)
+	if err != nil {
+		t.Fatalf("open pool: %v", err)
+	}
+
+	mux := http.NewServeMux()
+	authClient := auth.NewClient("https://test.invalid", "anon-key", []byte("test-secret"))
+	handlers.Register(mux, authClient, "", &handlers.Services{Pool: pool})
+
+	rec := httptest.NewRecorder()
+	req := httptest.NewRequest(http.MethodGet, "/health/ready", nil)
+	mux.ServeHTTP(rec, req)
+	if rec.Code != http.StatusOK {
+		t.Errorf("GET /health/ready (live pool): status=%d, body=%q; want 200", rec.Code, rec.Body.String())
+	}
+	if body := strings.TrimSpace(rec.Body.String()); body != "ready" {
+		t.Errorf("GET /health/ready (live pool): body=%q; want \"ready\"", body)
+	}
 }
 
 // embeddedMigrationVersions returns every N where N_*.up.sql exists in

docs/cicd-runner-setup-2026-05-25.md

@@ -0,0 +1,181 @@
+# CI/CD runner setup — paliad
+
+**Companion to:** `docs/design-cicd-pre-deploy-gate-2026-05-25.md` (Slice A, t-paliad-282 / m/paliad#114)
+**Date:** 2026-05-25
+**Audience:** mlake / mriver admin (m or head)
+
+Slice A's `.gitea/workflows/test.yaml` requires (a) at least one online Gitea Actions runner and (b) a Dokploy API token wired as a repo secret. Both are one-time setup actions that paliad's source tree cannot perform itself — they live on infra-side. This doc lists them so the workflow can go green on its first run.
+
+---
+
+## 0. Pre-flight: what already exists
+
+Verified live (2026-05-25 cronus inventor shift):
+
+- Gitea 1.24.4 on `mgit.msbls.de`, `has_actions: true` on `m/paliad`.
+- `/api/v1/admin/actions/runners` reports **2 runners** registered. They are likely the shared runners used by `m/mGreen` and `m/mGeo` (both have `.gitea/workflows/deploy.yml` with `runs-on: self-hosted`).
+- `m/paliad/actions/tasks` reports `total_count=0` — paliad has never run a workflow yet.
+
+The existing runners may already be capable of running paliad's workflow without further setup. The verification step (§3) below tells you whether they are.
+
+---
+
+## 1. Runner placement decision (m's Q11.1)
+
+m's pick: **mriver**.
+
+Rationale: mriver hosts the mai worker fleet but workers spend most of their time waiting on Anthropic. mlake's Dokploy + Swarm workload is more contended. A new runner on mriver adds the least pressure to either box.
+
+If mriver is offline or saturated when CI first fires, fall back to the existing mlake-side runners (they're already registered; no provisioning needed).
+
+---
+
+## 2. One-time setup (admin steps)
+
+### 2.1 Register a new Gitea Actions runner on mriver
+
+```bash
+# On mriver, as m:
+# 1. Download the act_runner binary (matching Gitea 1.24.x)
+curl -L -o /usr/local/bin/act_runner \
+  https://gitea.com/gitea/act_runner/releases/download/v0.2.13/act_runner-0.2.13-linux-amd64
+chmod +x /usr/local/bin/act_runner
+
+# 2. Get a runner registration token. In the Gitea UI:
+#    /admin → Actions → Runners → "Create new Runner"
+#    (or org-scope: /m/paliad/settings/actions/runners)
+# Copy the token.
+
+# 3. Register
+mkdir -p ~/act_runner && cd ~/act_runner
+act_runner register --no-interactive \
+  --instance https://mgit.msbls.de \
+  --token <REGISTRATION_TOKEN> \
+  --name mriver-paliad-1 \
+  --labels ubuntu-latest:docker://node:20-bookworm
+
+# 4. Run as a systemd unit (preferred) or as a session daemon
+# Systemd unit example: /etc/systemd/system/act_runner.service
+#   [Unit]
+#   Description=Gitea Actions runner
+#   After=network.target
+#   [Service]
+#   User=m
+#   WorkingDirectory=/home/m/act_runner
+#   ExecStart=/usr/local/bin/act_runner daemon
+#   Restart=on-failure
+#   [Install]
+#   WantedBy=multi-user.target
+sudo systemctl enable --now act_runner
+sudo systemctl status act_runner
+```
+
+**Why `ubuntu-latest:docker://node:20-bookworm` for the label?** Gitea Actions' `runs-on: ubuntu-latest` resolves via the runner's label map. Mapping it to a Docker image gives the workflow a sandbox with Docker available — required for our Postgres service container in `test.yaml`. mriver should have Docker (for `paliadin-shim`); if not, install it.
+
+### 2.2 Register the Dokploy API token as a repo secret
+
+The workflow's `deploy` job needs `secrets.DOKPLOY_TOKEN`. Use the existing project-wide Dokploy API key (the one stored in `~/.claude/skills/mai-dokploy/SKILL.md`).
+
+In the Gitea UI:
+- Navigate to `https://mgit.msbls.de/m/paliad/settings/actions/secrets`
+- Click "Add secret"
+  - Name: `DOKPLOY_TOKEN`
+  - Value: `mai-ottosSyRHMhmLhhhXaCbKzbqKBuSqzqEtmKDOPelPCeimTaYsbmaVslVyEgJZGCIxVdz`
+
+Or via API (mAi identity):
+```bash
+curl --netrc-file ~/.netrc-mai -sS -X POST \
+  -H "Content-Type: application/json" \
+  https://mgit.msbls.de/api/v1/repos/m/paliad/actions/secrets/DOKPLOY_TOKEN \
+  -d '{"data":"mai-ottosSyRHMhmLhhhXaCbKzbqKBuSqzqEtmKDOPelPCeimTaYsbmaVslVyEgJZGCIxVdz"}'
+```
+
+(Requires repo-owner permission. If mAi lacks it, m runs it.)
+
+---
+
+## 3. Verify the runner sees the workflow
+
+After (2.1) + (2.2):
+
+```bash
+# Push the Slice A branch (the one this doc lives on)
+git push origin mai/cronus/coder-cicd-slice-a
+
+# Confirm the runner picked up the job
+curl --netrc-file ~/.netrc-mai -sS \
+  "https://mgit.msbls.de/api/v1/repos/m/paliad/actions/tasks?limit=5" | jq '.'
+```
+
+A new task per job should appear (build, test-go). If `total_count` stays 0, the runner labels don't match the workflow's `runs-on`. Re-register with `--labels ubuntu-latest` (no docker:// suffix) and the existing runners on mlake will pick it up via shell mode.
+
+---
+
+## 4. Soft-launch (m's Q11.4)
+
+m's pick: **keep both Dokploy auto-deploy and the workflow's deploy step alive for ~1 week. After ≥5 successful green deploys via the workflow, disable Dokploy's autoDeploy in the Dokploy UI for the paliad compose.**
+
+While both are live, every push to main fires:
+1. Dokploy webhook (existing path) → deploys immediately, no gate.
+2. Gitea workflow → on green, ALSO calls `compose.deploy`.
+
+The second call is idempotent — if Dokploy already deployed the same commit, this is a no-op. The workflow's value during soft-launch is the **gate signal**: a red workflow on a green main = the bad migration shipped via the unguarded webhook and broke prod, and the workflow is shouting about it.
+
+After confidence builds:
+1. In the Dokploy UI, navigate to the paliad compose → Settings.
+2. Toggle "Auto Deploy" off.
+3. Save.
+
+From this point, the only path to deploy is the workflow's deploy job. Red workflow = no deploy.
+
+---
+
+## 5. What Slice A catches today — and what it doesn't
+
+After this branch (`mai/cronus/coder-cicd-slice-a`) merges to main:
+
+### Catches (active in CI)
+
+- **Build breakage** — `go build`, `go vet`, `bun run build`. Red gate, no deploy.
+- **Slot collisions** — `TestMigrations_NoDuplicateSlot` runs without a DB. A PR adding migration N when version N already exists fails at gate time. This is the brunel-class catch (m/paliad#114 ~13:20 outage).
+- **New-migration shape errors (hermes class)** — `TestBootSmoke` runs `ApplyMigrations` against the snapshot-restored DB. New migs from this PR get applied for real; any column/relation/syntax error fails the gate before merge.
+- **New-migration ownership errors (mig 129 42501 class)** — `TestMigrations_EndToEndAsAppRole` runs `ApplyMigrations` connected as `postgres` (NON-superuser on `supabase/postgres:15.8.1.060`, same role topology as youpc-supabase prod). Any migration that assumes supabase_admin privilege fails with the same `42501 must be owner` error class that took paliad.de offline on 2026-05-25.
+- **Readiness probe regressions** — `TestHealthReady_Live` confirms `/health/ready` returns 200 against a live pool, 503 against a nil pool.
+- **Pure-Go test regressions** — `go test ./internal/... ./cmd/...` runs without `TEST_DATABASE_URL` (live-DB service tests skip the same way they do on a developer laptop without a scratch DB).
+
+### Mechanism — the snapshot approach
+
+CI's scratch DB starts from a `pg_dump` of youpc-supabase paliad schema +
+`paliad.applied_migrations` rows, committed to `internal/db/testdata/prod-snapshot.sql`. After restore, the scratch DB is at "paliad HEAD of snapshot" and `ApplyMigrations` sees only this PR's new migrations as pending.
+
+This sidesteps the fresh-DB idempotence problem: several historical migrations (notably mig 037's missing `CREATE EXTENSION pg_trgm`, mig 051's inner `COMMIT;`) can't be replayed from scratch against `supabase/postgres:15.8.1.060`. The snapshot pins everything that's already applied in prod and lets CI focus on what's new — which is what we actually care about for outage prevention.
+
+Snapshot refresh: `make refresh-snapshot` with `PALIAD_PROD_DATABASE_URL` set (see `internal/db/testdata/README.md`).
+
+### Known gap — live-DB service tests don't run in CI
+
+`internal/services/*_test.go` tests with `TEST_DATABASE_URL` set fail against `supabase/postgres:15.8.1.060` with `42P08 inconsistent types deduced for parameter` errors on some INSERT bind paths. The same tests pass against youpc-supabase prod. Cause is unconfirmed — likely subtle differences in type inference between the dockerized image and the prod cluster's configuration. CI today runs `go test ./...` without `TEST_DATABASE_URL` so these tests skip. Not blocking outage prevention; tracked as a follow-up for the post-Slice-A coder.
+
+### Migration cleanup also bundled in this PR
+
+Two surgical migration improvements that surfaced during snapshot debugging — kept here because they're small and harmless:
+
+- **mig 024 + 027** — `ALTER INDEX` / `ALTER POLICY` exception handlers now catch `undefined_object` OR `undefined_table` OR `duplicate_object`. Old handler caught only `undefined_object`; Postgres raises `undefined_table` when the source object never existed and `duplicate_object` when the destination already exists. The expanded handler makes the migrations truly idempotent across the three plausible states: source-still-German (rename succeeds), already-renamed (catches duplicate_object), and fresh-DB-never-had-German (catches undefined_table).
+
+Other migration history bugs (mig 037 missing pg_trgm, mig 051 inner COMMIT) are tracked as a separate cleanup task — not blocking, because the snapshot bypasses them.
+
+### Verification checklist (after Slice A merges)
+
+1. **Workflow green on its first PR run?** Check `/m/paliad/actions`. If not, fix before merging.
+2. **Dokploy `compose.deploy` call succeeds?** The workflow's `deploy` job logs the POST response. A successful response is a Dokploy job ID; a 4xx is an auth or compose-id problem.
+3. **`/health/ready` returns 200 within 5 minutes after a green deploy?** The workflow polls this. If it times out, the migration may have failed silently inside the new container — check `docker logs --tail 50 compose-transmit-multi-byte-driver-v7jth9-web-1` on mlake.
+4. **Reproduce the slot-collision catch locally:** rename `131_…up.sql` to `129_…` (duplicate slot) → workflow MUST fail at `Migration coordination check`. Revert before pushing.
+5. **Reproduce the role-split catch locally:** add a no-op migration `132_test_supersedes.up.sql` containing `REINDEX SYSTEM paliad_scratch;` (requires superuser). Workflow MUST fail at `Migration end-to-end (deploy role)`. Revert before pushing.
+
+---
+
+## 6. Future polish (Slice D, m's Q4 R-pick)
+
+`mai-test` post-merge shift: once Slice A is stable, wire a Gitea webhook on push-to-main that fires `/mai-test` as a follow-up shift. It runs the broader smoke + integration suite and posts results as a Gitea commit status. Not blocking; the gate doesn't depend on it.
+
+Implementation belongs in `m/mAi` (the mai webhook handler), not in paliad. Out of scope for Slice A.

internal/db/migrate_test.go

@@ -116,6 +116,57 @@ func TestMigrations_DryRun(t *testing.T) {
 	}
 }
 
+// TestMigrations_NoDuplicateSlot is a free-standing pre-flight check that
+// scanEmbeddedMigrations refuses to walk a tree where two *.up.sql files
+// claim the same NNN slot. This is the brunel-slot-collision class of
+// outage (m/paliad#114, 2026-05-25 ~13:20): a worker writes a migration
+// at slot N while another shipped slot N from a separate branch, both
+// merge, both end up in the embed.FS, and the runner refuses to start.
+//
+// Catching this at CI time (no DB needed) lets the second PR fail before
+// it merges, instead of breaking prod at the next deploy. Pure unit test;
+// runs even on developer laptops that don't set TEST_DATABASE_URL.
+func TestMigrations_NoDuplicateSlot(t *testing.T) {
+	if _, err := scanEmbeddedMigrations(); err != nil {
+		t.Fatalf("scanEmbeddedMigrations: %v "+
+			"(two migrations share the same NNN slot — coordinate with head "+
+			"and rename one of them before merging)", err)
+	}
+}
+
+// TestMigrations_EndToEndAsAppRole applies every embedded migration in
+// numeric order against a scratch DB connected as a NON-SUPERUSER role.
+// This is the prod-shape smoke that the per-mig BEGIN/ROLLBACK dry-run
+// (TestMigrations_DryRun) cannot deliver: the dry-run runs each
+// statement in isolation and rolls back, so it cannot reproduce the
+// mig-129-class outage (m/paliad#114, 2026-05-25 ~14:56 — pq: must be
+// owner of table project_event_choices, SQLSTATE 42501) where a
+// migration assumes ownership the deploy role doesn't have.
+//
+// Requires TEST_APP_DATABASE_URL — a Postgres URL whose role is NOT a
+// superuser and does NOT own the `paliad` schema (m's Q11.2 pick:
+// generic two-role model, see docs/design-cicd-pre-deploy-gate-2026-05-25.md
+// §6.2(a)). The CI workflow creates the role + schema split before
+// invoking the test; a developer who wants to reproduce the gate locally
+// runs the same SQL preamble (see Makefile target `verify-migrations`).
+//
+// Skipped without TEST_APP_DATABASE_URL — keeps `go test ./...` green
+// on machines that haven't set up the role split.
+func TestMigrations_EndToEndAsAppRole(t *testing.T) {
+	url := os.Getenv("TEST_APP_DATABASE_URL")
+	if url == "" {
+		t.Skip("TEST_APP_DATABASE_URL not set — skipping role-split end-to-end migration smoke")
+	}
+	if err := ApplyMigrations(url); err != nil {
+		t.Fatalf("ApplyMigrations as app role failed: %v "+
+			"(a migration assumes more privilege than the deploy role has — "+
+			"common cases: ALTER TABLE on a schema-owner table, CREATE EXTENSION "+
+			"without grants, SET ROLE without permission. Fix the migration to "+
+			"work as the deploy role, or arrange for the schema to be owned by "+
+			"the deploy role)", err)
+	}
+}
+
 // readAppliedVersions returns the set of versions present in
 // paliad.applied_migrations on the scratch DB. Missing table → empty set
 // (fresh-DB path; the table only exists after the runner has been called).

internal/db/migrations/024_rename_department_columns.up.sql

@@ -26,24 +26,24 @@ DO $$ BEGIN ALTER TABLE paliad.department_members RENAME COLUMN dezernat_id TO d
 -- Constraints (primary key + foreign keys + check). Renaming a pkey
 -- constraint also renames the underlying index of the same name.
 -- ---------------------------------------------------------------------------
-DO $$ BEGIN ALTER TABLE paliad.departments        RENAME CONSTRAINT dezernate_pkey                          TO departments_pkey;                       EXCEPTION WHEN undefined_object THEN NULL; END $$;
+DO $$ BEGIN ALTER TABLE paliad.departments        RENAME CONSTRAINT dezernate_pkey                          TO departments_pkey;                       EXCEPTION WHEN undefined_object OR undefined_table OR duplicate_object THEN NULL; END $$;
-DO $$ BEGIN ALTER TABLE paliad.departments        RENAME CONSTRAINT dezernate_lead_user_id_fkey             TO departments_lead_user_id_fkey;          EXCEPTION WHEN undefined_object THEN NULL; END $$;
+DO $$ BEGIN ALTER TABLE paliad.departments        RENAME CONSTRAINT dezernate_lead_user_id_fkey             TO departments_lead_user_id_fkey;          EXCEPTION WHEN undefined_object OR undefined_table OR duplicate_object THEN NULL; END $$;
-DO $$ BEGIN ALTER TABLE paliad.departments        RENAME CONSTRAINT dezernate_office_check                  TO departments_office_check;               EXCEPTION WHEN undefined_object THEN NULL; END $$;
+DO $$ BEGIN ALTER TABLE paliad.departments        RENAME CONSTRAINT dezernate_office_check                  TO departments_office_check;               EXCEPTION WHEN undefined_object OR undefined_table OR duplicate_object THEN NULL; END $$;
-DO $$ BEGIN ALTER TABLE paliad.department_members RENAME CONSTRAINT dezernat_mitglieder_pkey                 TO department_members_pkey;                EXCEPTION WHEN undefined_object THEN NULL; END $$;
+DO $$ BEGIN ALTER TABLE paliad.department_members RENAME CONSTRAINT dezernat_mitglieder_pkey                 TO department_members_pkey;                EXCEPTION WHEN undefined_object OR undefined_table OR duplicate_object THEN NULL; END $$;
-DO $$ BEGIN ALTER TABLE paliad.department_members RENAME CONSTRAINT dezernat_mitglieder_dezernat_id_fkey     TO department_members_department_id_fkey;  EXCEPTION WHEN undefined_object THEN NULL; END $$;
+DO $$ BEGIN ALTER TABLE paliad.department_members RENAME CONSTRAINT dezernat_mitglieder_dezernat_id_fkey     TO department_members_department_id_fkey;  EXCEPTION WHEN undefined_object OR undefined_table OR duplicate_object THEN NULL; END $$;
-DO $$ BEGIN ALTER TABLE paliad.department_members RENAME CONSTRAINT dezernat_mitglieder_user_id_fkey         TO department_members_user_id_fkey;        EXCEPTION WHEN undefined_object THEN NULL; END $$;
+DO $$ BEGIN ALTER TABLE paliad.department_members RENAME CONSTRAINT dezernat_mitglieder_user_id_fkey         TO department_members_user_id_fkey;        EXCEPTION WHEN undefined_object OR undefined_table OR duplicate_object THEN NULL; END $$;
 
 -- ---------------------------------------------------------------------------
 -- Standalone indexes (non-pkey).
 -- ---------------------------------------------------------------------------
-DO $$ BEGIN ALTER INDEX paliad.dezernate_office_idx           RENAME TO departments_office_idx;        EXCEPTION WHEN undefined_object THEN NULL; END $$;
+DO $$ BEGIN ALTER INDEX paliad.dezernate_office_idx           RENAME TO departments_office_idx;        EXCEPTION WHEN undefined_object OR undefined_table OR duplicate_object THEN NULL; END $$;
-DO $$ BEGIN ALTER INDEX paliad.dezernate_lead_idx             RENAME TO departments_lead_idx;          EXCEPTION WHEN undefined_object THEN NULL; END $$;
+DO $$ BEGIN ALTER INDEX paliad.dezernate_lead_idx             RENAME TO departments_lead_idx;          EXCEPTION WHEN undefined_object OR undefined_table OR duplicate_object THEN NULL; END $$;
-DO $$ BEGIN ALTER INDEX paliad.dezernat_mitglieder_user_idx   RENAME TO department_members_user_idx;   EXCEPTION WHEN undefined_object THEN NULL; END $$;
+DO $$ BEGIN ALTER INDEX paliad.dezernat_mitglieder_user_idx   RENAME TO department_members_user_idx;   EXCEPTION WHEN undefined_object OR undefined_table OR duplicate_object THEN NULL; END $$;
 
 -- ---------------------------------------------------------------------------
 -- RLS policies
 -- ---------------------------------------------------------------------------
-DO $$ BEGIN ALTER POLICY dezernate_select           ON paliad.departments        RENAME TO departments_select;        EXCEPTION WHEN undefined_object THEN NULL; END $$;
+DO $$ BEGIN ALTER POLICY dezernate_select           ON paliad.departments        RENAME TO departments_select;        EXCEPTION WHEN undefined_object OR undefined_table OR duplicate_object THEN NULL; END $$;
-DO $$ BEGIN ALTER POLICY dezernate_write            ON paliad.departments        RENAME TO departments_write;         EXCEPTION WHEN undefined_object THEN NULL; END $$;
+DO $$ BEGIN ALTER POLICY dezernate_write            ON paliad.departments        RENAME TO departments_write;         EXCEPTION WHEN undefined_object OR undefined_table OR duplicate_object THEN NULL; END $$;
-DO $$ BEGIN ALTER POLICY dezernat_mitglieder_select ON paliad.department_members RENAME TO department_members_select; EXCEPTION WHEN undefined_object THEN NULL; END $$;
+DO $$ BEGIN ALTER POLICY dezernat_mitglieder_select ON paliad.department_members RENAME TO department_members_select; EXCEPTION WHEN undefined_object OR undefined_table OR duplicate_object THEN NULL; END $$;
-DO $$ BEGIN ALTER POLICY dezernat_mitglieder_write  ON paliad.department_members RENAME TO department_members_write;  EXCEPTION WHEN undefined_object THEN NULL; END $$;
+DO $$ BEGIN ALTER POLICY dezernat_mitglieder_write  ON paliad.department_members RENAME TO department_members_write;  EXCEPTION WHEN undefined_object OR undefined_table OR duplicate_object THEN NULL; END $$;

internal/db/migrations/027_rename_to_partner_units.up.sql

@@ -63,27 +63,27 @@ ALTER TABLE paliad.partner_unit_members RENAME COLUMN department_id TO partner_u
 -- 5. Rename constraints. Postgres auto-renames the underlying index for
 --    pkey/uniq constraints; standalone indexes are renamed in step 6.
 -- ---------------------------------------------------------------------------
-DO $$ BEGIN ALTER TABLE paliad.partner_units        RENAME CONSTRAINT departments_pkey                       TO partner_units_pkey;                             EXCEPTION WHEN undefined_object THEN NULL; END $$;
+DO $$ BEGIN ALTER TABLE paliad.partner_units        RENAME CONSTRAINT departments_pkey                       TO partner_units_pkey;                             EXCEPTION WHEN undefined_object OR undefined_table OR duplicate_object THEN NULL; END $$;
-DO $$ BEGIN ALTER TABLE paliad.partner_units        RENAME CONSTRAINT departments_lead_user_id_fkey          TO partner_units_lead_user_id_fkey;                EXCEPTION WHEN undefined_object THEN NULL; END $$;
+DO $$ BEGIN ALTER TABLE paliad.partner_units        RENAME CONSTRAINT departments_lead_user_id_fkey          TO partner_units_lead_user_id_fkey;                EXCEPTION WHEN undefined_object OR undefined_table OR duplicate_object THEN NULL; END $$;
-DO $$ BEGIN ALTER TABLE paliad.partner_units        RENAME CONSTRAINT departments_office_check               TO partner_units_office_check;                     EXCEPTION WHEN undefined_object THEN NULL; END $$;
+DO $$ BEGIN ALTER TABLE paliad.partner_units        RENAME CONSTRAINT departments_office_check               TO partner_units_office_check;                     EXCEPTION WHEN undefined_object OR undefined_table OR duplicate_object THEN NULL; END $$;
-DO $$ BEGIN ALTER TABLE paliad.partner_unit_members RENAME CONSTRAINT department_members_pkey                TO partner_unit_members_pkey;                      EXCEPTION WHEN undefined_object THEN NULL; END $$;
+DO $$ BEGIN ALTER TABLE paliad.partner_unit_members RENAME CONSTRAINT department_members_pkey                TO partner_unit_members_pkey;                      EXCEPTION WHEN undefined_object OR undefined_table OR duplicate_object THEN NULL; END $$;
-DO $$ BEGIN ALTER TABLE paliad.partner_unit_members RENAME CONSTRAINT department_members_department_id_fkey  TO partner_unit_members_partner_unit_id_fkey;      EXCEPTION WHEN undefined_object THEN NULL; END $$;
+DO $$ BEGIN ALTER TABLE paliad.partner_unit_members RENAME CONSTRAINT department_members_department_id_fkey  TO partner_unit_members_partner_unit_id_fkey;      EXCEPTION WHEN undefined_object OR undefined_table OR duplicate_object THEN NULL; END $$;
-DO $$ BEGIN ALTER TABLE paliad.partner_unit_members RENAME CONSTRAINT department_members_user_id_fkey        TO partner_unit_members_user_id_fkey;              EXCEPTION WHEN undefined_object THEN NULL; END $$;
+DO $$ BEGIN ALTER TABLE paliad.partner_unit_members RENAME CONSTRAINT department_members_user_id_fkey        TO partner_unit_members_user_id_fkey;              EXCEPTION WHEN undefined_object OR undefined_table OR duplicate_object THEN NULL; END $$;
 
 -- ---------------------------------------------------------------------------
 -- 6. Rename non-pkey indexes.
 -- ---------------------------------------------------------------------------
-DO $$ BEGIN ALTER INDEX paliad.departments_office_idx        RENAME TO partner_units_office_idx;        EXCEPTION WHEN undefined_object THEN NULL; END $$;
+DO $$ BEGIN ALTER INDEX paliad.departments_office_idx        RENAME TO partner_units_office_idx;        EXCEPTION WHEN undefined_object OR undefined_table OR duplicate_object THEN NULL; END $$;
-DO $$ BEGIN ALTER INDEX paliad.departments_lead_idx          RENAME TO partner_units_lead_idx;          EXCEPTION WHEN undefined_object THEN NULL; END $$;
+DO $$ BEGIN ALTER INDEX paliad.departments_lead_idx          RENAME TO partner_units_lead_idx;          EXCEPTION WHEN undefined_object OR undefined_table OR duplicate_object THEN NULL; END $$;
-DO $$ BEGIN ALTER INDEX paliad.department_members_user_idx   RENAME TO partner_unit_members_user_idx;   EXCEPTION WHEN undefined_object THEN NULL; END $$;
+DO $$ BEGIN ALTER INDEX paliad.department_members_user_idx   RENAME TO partner_unit_members_user_idx;   EXCEPTION WHEN undefined_object OR undefined_table OR duplicate_object THEN NULL; END $$;
 
 -- ---------------------------------------------------------------------------
 -- 7. Rename RLS policies.
 -- ---------------------------------------------------------------------------
-DO $$ BEGIN ALTER POLICY departments_select        ON paliad.partner_units        RENAME TO partner_units_select;        EXCEPTION WHEN undefined_object THEN NULL; END $$;
+DO $$ BEGIN ALTER POLICY departments_select        ON paliad.partner_units        RENAME TO partner_units_select;        EXCEPTION WHEN undefined_object OR undefined_table OR duplicate_object THEN NULL; END $$;
-DO $$ BEGIN ALTER POLICY departments_write         ON paliad.partner_units        RENAME TO partner_units_write;         EXCEPTION WHEN undefined_object THEN NULL; END $$;
+DO $$ BEGIN ALTER POLICY departments_write         ON paliad.partner_units        RENAME TO partner_units_write;         EXCEPTION WHEN undefined_object OR undefined_table OR duplicate_object THEN NULL; END $$;
-DO $$ BEGIN ALTER POLICY department_members_select ON paliad.partner_unit_members RENAME TO partner_unit_members_select; EXCEPTION WHEN undefined_object THEN NULL; END $$;
+DO $$ BEGIN ALTER POLICY department_members_select ON paliad.partner_unit_members RENAME TO partner_unit_members_select; EXCEPTION WHEN undefined_object OR undefined_table OR duplicate_object THEN NULL; END $$;
-DO $$ BEGIN ALTER POLICY department_members_write  ON paliad.partner_unit_members RENAME TO partner_unit_members_write;  EXCEPTION WHEN undefined_object THEN NULL; END $$;
+DO $$ BEGIN ALTER POLICY department_members_write  ON paliad.partner_unit_members RENAME TO partner_unit_members_write;  EXCEPTION WHEN undefined_object OR undefined_table OR duplicate_object THEN NULL; END $$;
 
 -- ---------------------------------------------------------------------------
 -- 8. Audit table for partner-unit events. Mutations on partner_units +

internal/db/testdata/README.md

@@ -0,0 +1,69 @@
+# `internal/db/testdata/` — CI snapshot
+
+## `prod-snapshot.sql`
+
+Schema-only `pg_dump` of paliad's prod DB (youpc-supabase paliad schema)
+plus the rows of `paliad.applied_migrations` that match this branch's
+on-disk migration set.
+
+**Purpose.** Lets CI's migration smoke (`.gitea/workflows/test.yaml`)
+restore a Postgres scratch DB to "paliad at HEAD-of-snapshot" without
+having to replay 131 migrations from scratch. ApplyMigrations on the
+restored DB sees the applied set and only runs whatever NEW migrations
+this PR adds — exactly the integration shape we want to test, and the
+same shape prod sees on every deploy.
+
+**Why a snapshot at all.** Running ApplyMigrations from scratch against a
+fresh `supabase/postgres:15.8.1.060` surfaces multiple fresh-DB
+idempotence bugs in historical migrations (raw `COMMIT;` in mig 051,
+missing `CREATE EXTENSION pg_trgm` for mig 037, ALTER POLICY
+exception-handler gaps in mig 024/027 — the last is fixed in this PR).
+Fixing them all is a separate cleanup. The snapshot sidesteps them by
+starting CI from a state where every historical migration is already
+applied as it was in prod.
+
+**Schema scope.** `--schema=paliad` only. Auth schema comes baked into
+`supabase/postgres`; CI's setup step installs `pg_trgm` before restoring.
+
+**Ownership.** `--no-owner --no-privileges` keeps the dump portable
+across role topologies (CI's supabase_admin / postgres / authenticated /
+anon don't have to match prod's exact role layout). The role-split smoke
+relies on `postgres` being a non-superuser, which is true on
+supabase/postgres by default.
+
+**Refresh.** Run `make refresh-snapshot` with `PALIAD_PROD_DATABASE_URL`
+set to a Postgres URL with `pg_dump` rights on youpc-supabase. The
+target appends data rows for `paliad.applied_migrations`, strips
+`\restrict` / `\unrestrict` commands (pg 16 dump → pg 15 restore), and
+filters out applied-migrations rows for versions beyond the branch's
+local max. The CI workflow consumes the resulting file verbatim.
+
+**Verify a refresh.** Boot a local scratch:
+
+```bash
+docker run -d --rm --name paliad-snap \
+  -e POSTGRES_PASSWORD=ci -e POSTGRES_DB=paliad_scratch \
+  -p 15433:5432 supabase/postgres:15.8.1.060
+sleep 5
+docker exec -e PGPASSWORD=ci paliad-snap psql -h localhost -U supabase_admin -d paliad_scratch \
+  -c "GRANT CREATE ON DATABASE paliad_scratch TO postgres;" \
+  -c "CREATE EXTENSION IF NOT EXISTS pg_trgm;"
+cat internal/db/testdata/prod-snapshot.sql | docker exec -i -e PGPASSWORD=ci paliad-snap \
+  psql -h localhost -U postgres -d paliad_scratch -v ON_ERROR_STOP=1
+TEST_DATABASE_URL="postgres://postgres:ci@localhost:15433/paliad_scratch?sslmode=disable" \
+TEST_APP_DATABASE_URL="postgres://postgres:ci@localhost:15433/paliad_scratch?sslmode=disable" \
+  go test -count=1 -run 'TestMigrations|TestBootSmoke|TestHealthReady_Live' ./internal/db/ ./cmd/server/
+docker stop paliad-snap
+```
+
+All four named tests must pass. If any fails after a refresh,
+investigate before merging — usually because a new migration was added
+to prod that this branch doesn't have on disk yet.
+
+**Why is the snapshot not gzipped?** Small enough (~200 KB) that the
+diff stays human-readable in `git diff` reviews. If it crosses ~1 MB,
+gzip + decompress-on-restore in CI.
+
+**Privacy.** Schema-only dump, no row data from any paliad table (except
+`paliad.applied_migrations`, which contains migration filenames +
+checksums — public info already in the repo).

internal/handlers/handlers.go

@@ -1,9 +1,13 @@
 package handlers
 
 import (
+	"context"
 	"encoding/json"
 	"net/http"
 	"strings"
+	"time"
+
+	"github.com/jmoiron/sqlx"
 
 	"mgit.msbls.de/m/paliad/internal/auth"
 	"mgit.msbls.de/m/paliad/internal/services"
@@ -50,6 +54,12 @@ func noCachePages(h http.Handler) http.Handler {
 // Services bundles the Phase B + C database-backed services. Pass nil if
 // DATABASE_URL was unset; the matter-management endpoints will return 503.
 type Services struct {
+	// Pool is the raw connection pool. Held so the readiness probe
+	// (/health/ready) can ping it without going through any individual
+	// service. nil when DATABASE_URL was unset — in that case
+	// /health/ready returns 503.
+	Pool *sqlx.DB
+
 	Project        *services.ProjectService
 	Team           *services.TeamService
 	PartnerUnit    *services.PartnerUnitService
@@ -188,6 +198,38 @@ func Register(mux *http.ServeMux, client *auth.Client, giteaAPIToken string, svc
 		_, _ = w.Write([]byte("ok\n"))
 	})
 
+	// Readiness probe. Public, no auth. Distinct from /healthz: this
+	// returns 200 only when the DB pool is reachable. Reaching Register
+	// at all implies db.ApplyMigrations succeeded (cmd/server/main.go
+	// calls it before constructing svc), so a 200 here means "migrations
+	// applied AND pool responsive" — the contract Dokploy / Traefik should
+	// gate on, not the bind-and-serve check that /healthz answers.
+	//
+	// Three outcomes:
+	//   - svc == nil OR svc.Pool == nil  → 503 (DB-less knowledge-platform
+	//     deployments report not-ready so an external orchestrator can
+	//     distinguish them from a full prod boot).
+	//   - PingContext fails within 2 s   → 503 (pool unreachable).
+	//   - PingContext succeeds            → 200 "ready".
+	//
+	// Used by docker-compose.yml's healthcheck (Slice B) and by the
+	// post-deploy verification step in .gitea/workflows/test.yaml.
+	mux.HandleFunc("GET /health/ready", func(w http.ResponseWriter, r *http.Request) {
+		w.Header().Set("Cache-Control", "no-store")
+		w.Header().Set("Content-Type", "text/plain; charset=utf-8")
+		if svc == nil || svc.Pool == nil {
+			http.Error(w, "db not configured\n", http.StatusServiceUnavailable)
+			return
+		}
+		ctx, cancel := context.WithTimeout(r.Context(), 2*time.Second)
+		defer cancel()
+		if err := svc.Pool.PingContext(ctx); err != nil {
+			http.Error(w, "db unreachable\n", http.StatusServiceUnavailable)
+			return
+		}
+		_, _ = w.Write([]byte("ready\n"))
+	})
+
 	// API endpoints (JSON, public)
 	mux.HandleFunc("POST /api/login", handleAPILogin)
 	mux.HandleFunc("POST /api/register", handleAPIRegister)