ci: full test gate — types, frontend build, agent tests, agent<->backend contract suite

ci.yml runs on every push to main: backend tsc, frontend vue-tsc+vite,
cargo test (cached), then an integration job with postgres:16 + nats
service containers — real migrations applied to a fresh DB, real
backend booted (admin seed provides the license), real agent binary
spawned. contract-tests/agent-backend.contract.mjs proves the entire
v2 pipeline: heartbeat shape + measured telemetry, auto-registered
server_connections row flipping connected, instance start/stop/status
round-trips with push events, and the offline beacon flipping the row
back. This is the test that could not run before a production rebuild
until now — it now runs before every push lands.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

.gitea/workflows/ci.yml

@@ -0,0 +1,122 @@
+name: CI
+
+# Test gate for every push to main. The deploy story: main must be green here
+# before the stack is rebuilt (deploy workflow enforces it once SSH transport
+# secrets land). Jobs run in the runner's bare node:20-bullseye container —
+# toolchains bootstrap per-run.
+
+on:
+  push:
+    branches: [main]
+  pull_request:
+
+jobs:
+  backend-types:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - name: Type-check NestJS backend
+        run: |
+          cd backend-nest
+          npm ci --no-audit --no-fund 2>&1 | tail -2
+          npx tsc --noEmit
+
+  frontend-build:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - name: Build frontend (vue-tsc gate + vite)
+        run: |
+          cd frontend
+          npm ci --no-audit --no-fund 2>&1 | tail -2
+          npm run build
+
+  agent-tests:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - name: Cache cargo
+        uses: actions/cache@v4
+        with:
+          path: |
+            ~/.cargo/registry
+            ~/.cargo/git
+            corrosion-host-agent/target
+          key: cargo-${{ hashFiles('corrosion-host-agent/Cargo.lock') }}
+      - name: Install Rust
+        run: |
+          apt-get update -qq && apt-get install -y -qq build-essential curl
+          curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y --default-toolchain stable --profile minimal
+          echo "$HOME/.cargo/bin" >> $GITHUB_PATH
+      - name: Test agent
+        run: |
+          cd corrosion-host-agent
+          cargo test
+      - name: Upload agent binary for integration
+        uses: actions/upload-artifact@v3
+        with:
+          name: agent-debug
+          path: corrosion-host-agent/target/debug/corrosion-host-agent
+
+  integration:
+    runs-on: ubuntu-latest
+    needs: agent-tests
+    services:
+      postgres:
+        image: postgres:16
+        env:
+          POSTGRES_USER: corrosion
+          POSTGRES_PASSWORD: citest
+          POSTGRES_DB: corrosion
+      nats:
+        image: nats:2.10-alpine
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Download agent binary
+        uses: actions/download-artifact@v3
+        with:
+          name: agent-debug
+          path: agent-bin
+
+      - name: Apply migrations to fresh DB
+        run: |
+          apt-get update -qq && apt-get install -y -qq postgresql-client
+          until PGPASSWORD=citest psql -h postgres -U corrosion -d corrosion -c 'SELECT 1' >/dev/null 2>&1; do sleep 1; done
+          for f in $(ls backend/migrations/*.sql | sort); do
+            echo "applying $f"
+            PGPASSWORD=citest psql -h postgres -U corrosion -d corrosion -v ON_ERROR_STOP=1 -q -f "$f"
+          done
+
+      - name: Build + boot backend
+        run: |
+          cd backend-nest
+          npm ci --no-audit --no-fund 2>&1 | tail -2
+          npm run build
+          DATABASE_URL=postgres://corrosion:citest@postgres:5432/corrosion \
+          NATS_URL=nats://nats:4222 \
+          JWT_SECRET=ci-secret ENCRYPTION_KEY=ci-encryption-key \
+          ADMIN_EMAIL=ci@corrosion.test ADMIN_PASSWORD=ci-password-123 ADMIN_USERNAME=CI \
+          nohup node dist/main.js > /tmp/backend.log 2>&1 &
+          for i in $(seq 1 30); do
+            code=$(curl -s -o /dev/null -w "%{http_code}" http://localhost:3000/api/auth/login -X POST -H 'Content-Type: application/json' -d '{}' || true)
+            [ "$code" = "400" ] && echo "backend up" && exit 0
+            sleep 2
+          done
+          echo "backend failed to come up"; cat /tmp/backend.log; exit 1
+
+      - name: Run agent↔backend contract suite
+        run: |
+          chmod +x agent-bin/corrosion-host-agent
+          LICENSE_ID=$(PGPASSWORD=citest psql -h postgres -U corrosion -d corrosion -t -A -c 'SELECT id FROM licenses LIMIT 1')
+          echo "license under test: $LICENSE_ID"
+          [ -n "$LICENSE_ID" ] || { echo "admin seed did not create a license"; cat /tmp/backend.log; exit 1; }
+          LICENSE_ID="$LICENSE_ID" \
+          DATABASE_URL=postgres://corrosion:citest@postgres:5432/corrosion \
+          NATS_URL=nats://nats:4222 \
+          AGENT_BIN=$PWD/agent-bin/corrosion-host-agent \
+          node contract-tests/agent-backend.contract.mjs
+
+      - name: Backend log on failure
+        if: failure()
+        run: cat /tmp/backend.log || true

contract-tests/agent-backend.contract.mjs

@@ -0,0 +1,152 @@
+// Full-pipeline contract test: Rust host agent → NATS → NestJS consumer → Postgres.
+//
+// Proves the wire protocol v2 chain end to end against a REAL backend and DB:
+//   1. agent heartbeat arrives with schema 2 + measured telemetry
+//   2. backend auto-registers the server_connections row and marks it connected
+//   3. instance command channel round-trips (start/status/stop) with push events
+//   4. graceful agent shutdown publishes the offline beacon and the row flips offline
+//
+// Required env:
+//   LICENSE_ID    — existing license uuid (CI: from the admin seed)
+//   DATABASE_URL  — postgres connection string for assertions
+//   NATS_URL      — broker both agent and backend use (default nats://localhost:4222)
+//   AGENT_BIN     — path to the corrosion-host-agent binary
+//
+// Uses the backend's own node_modules (nats, pg) so the client libs under test
+// are exactly what production runs.
+
+import { createRequire } from 'node:module';
+import { spawn } from 'node:child_process';
+import { writeFileSync, mkdtempSync } from 'node:fs';
+import { tmpdir } from 'node:os';
+import { join, dirname } from 'node:path';
+import { fileURLToPath } from 'node:url';
+
+const repoRoot = join(dirname(fileURLToPath(import.meta.url)), '..');
+const require = createRequire(join(repoRoot, 'backend-nest', 'node_modules', 'x.js'));
+const { connect, StringCodec } = require('nats');
+const { Client: PgClient } = require('pg');
+
+const LICENSE = process.env.LICENSE_ID;
+const NATS_URL = process.env.NATS_URL ?? 'nats://localhost:4222';
+const DATABASE_URL = process.env.DATABASE_URL;
+const AGENT_BIN = process.env.AGENT_BIN ?? join(repoRoot, 'corrosion-host-agent', 'target', 'debug', 'corrosion-host-agent');
+
+if (!LICENSE || !DATABASE_URL) {
+  console.error('LICENSE_ID and DATABASE_URL are required');
+  process.exit(2);
+}
+
+const sc = StringCodec();
+const errs = [];
+const check = (cond, msg) => { if (!cond) errs.push(msg); };
+const sleep = (ms) => new Promise((r) => setTimeout(r, ms));
+
+async function pollDb(pg, predicate, label, timeoutMs = 30_000) {
+  const deadline = Date.now() + timeoutMs;
+  for (;;) {
+    const { rows } = await pg.query(
+      'SELECT connection_type, connection_status, companion_last_seen FROM server_connections WHERE license_id = $1',
+      [LICENSE],
+    );
+    if (predicate(rows)) return rows;
+    if (Date.now() > deadline) {
+      errs.push(`${label}: timeout after ${timeoutMs}ms — rows: ${JSON.stringify(rows)}`);
+      return rows;
+    }
+    await sleep(1000);
+  }
+}
+
+const main = async () => {
+  const pg = new PgClient({ connectionString: DATABASE_URL });
+  await pg.connect();
+  const nc = await connect({ servers: NATS_URL });
+
+  const heartbeats = [];
+  const statusEvents = [];
+  (async () => { for await (const m of nc.subscribe(`corrosion.${LICENSE}.host.heartbeat`)) heartbeats.push(JSON.parse(sc.decode(m.data))); })();
+  (async () => { for await (const m of nc.subscribe(`corrosion.${LICENSE}.ci-instance.status`)) statusEvents.push(JSON.parse(sc.decode(m.data))); })();
+
+  // --- spawn the real agent ---
+  const dir = mkdtempSync(join(tmpdir(), 'cha-contract-'));
+  const cfgPath = join(dir, 'agent.toml');
+  writeFileSync(cfgPath, `
+[agent]
+license_id = "${LICENSE}"
+nats_url = "${NATS_URL}"
+heartbeat_seconds = 10
+log_level = "info"
+
+[[instance]]
+id = "ci-instance"
+game = "rust"
+root = "/tmp"
+label = "Contract CI"
+executable = "/bin/sleep"
+args = ["300"]
+`);
+  const agent = spawn(AGENT_BIN, ['--config', cfgPath], { stdio: ['ignore', 'inherit', 'inherit'] });
+  const agentExited = new Promise((r) => agent.on('exit', r));
+
+  // --- 1. heartbeat shape + real telemetry ---
+  const hbDeadline = Date.now() + 20_000;
+  while (heartbeats.length === 0 && Date.now() < hbDeadline) await sleep(500);
+  check(heartbeats.length > 0, 'no heartbeat within 20s');
+  if (heartbeats.length) {
+    const hb = heartbeats[0];
+    check(hb.schema === 2, `schema != 2: ${hb.schema}`);
+    check(typeof hb.host?.cpu_percent === 'number', 'missing host.cpu_percent');
+    check(hb.host?.mem_total_mb > 0, 'mem_total_mb not measured');
+    check(Array.isArray(hb.host?.disks) && hb.host.disks.length > 0, 'no disks reported');
+    check(hb.instances?.[0]?.id === 'ci-instance', 'instance missing from heartbeat');
+    check(!!hb.agent?.version && !!hb.agent?.commit, 'agent version/commit missing');
+  }
+
+  // --- 2. backend auto-registers + connects ---
+  const rows = await pollDb(pg, (r) => r.length === 1 && r[0].connection_status === 'connected', 'auto-register connected');
+  if (rows.length === 1) {
+    check(rows[0].connection_type === 'bare_metal', `connection_type: ${rows[0].connection_type}`);
+    check(rows[0].companion_last_seen !== null, 'companion_last_seen not set');
+  }
+
+  // --- 3. instance command channel ---
+  const cmd = async (payload) =>
+    JSON.parse(sc.decode((await nc.request(`corrosion.${LICENSE}.ci-instance.cmd`, sc.encode(JSON.stringify(payload)), { timeout: 8000 })).data));
+
+  const st0 = await cmd({ func: 'status' });
+  check(st0.state?.state === 'stopped', `initial state: ${JSON.stringify(st0.state)}`);
+  const start = await cmd({ func: 'start' });
+  check(start.status === 'success', `start: ${JSON.stringify(start)}`);
+  await sleep(1000);
+  const st1 = await cmd({ func: 'status' });
+  check(st1.state?.state === 'running', `post-start state: ${JSON.stringify(st1.state)}`);
+  check((await cmd({ func: 'start' })).status === 'error', 'double start must error');
+  check((await cmd({ func: 'bogus' })).status === 'error', 'unknown func must error');
+  const stop = await cmd({ func: 'stop' });
+  check(stop.status === 'success', `stop: ${JSON.stringify(stop)}`);
+  await sleep(1000);
+  const seq = statusEvents.map((e) => e.event?.state);
+  check(seq.includes('running') && seq.includes('stopped'), `status events incomplete: ${seq.join(',')}`);
+
+  // --- 4. graceful shutdown → offline beacon → DB flips offline ---
+  agent.kill('SIGTERM');
+  await Promise.race([agentExited, sleep(8000)]);
+  await pollDb(pg, (r) => r.length === 1 && r[0].connection_status === 'offline', 'beacon offline', 20_000);
+
+  await nc.close();
+  await pg.end();
+
+  if (errs.length) {
+    console.error('\nCONTRACT FAIL:');
+    errs.forEach((e) => console.error(' -', e));
+    process.exit(1);
+  }
+  console.log('\nCONTRACT PASS: heartbeat shape, auto-register, connected/offline lifecycle, instance command channel, push events');
+  process.exit(0);
+};
+
+main().catch((e) => {
+  console.error('contract test crashed:', e);
+  process.exit(1);
+});