feat(federation): Step-CA client service for grant certs (FED-M2-04)

- Add CaService (@Injectable) that POSTs CSRs to step-ca /1.0/sign over
  HTTPS with a pinned CA root cert; builds HS256 OTT with custom claims
  mosaic_grant_id and mosaic_subject_user_id plus step.sha CSR fingerprint
- Add CaServiceError with cause + remediation for fail-loud contract
- Add IssueCertRequestDto and IssuedCertDto with class-validator decorators
- Add FederationModule exporting CaService; wire into AppModule
- Replace federation.tpl TODO placeholder with real step-ca Go template
  emitting OID 1.3.6.1.4.1.99999.1 (grantId) and .2 (subjectUserId) as
  DER UTF8String extensions (tag 0x0C, length 0x24, base64-encoded value)
- Update infra/step-ca/init.sh to patch mosaic-fed provisioner config with
  templateFile path via jq on first boot (idempotent)
- Append OID assignment registry and CA env var table to docs/federation/SETUP.md
- 11 unit tests pass: happy path, certChain fallbacks, HTTP 401/4xx, malformed
  CSR (no HTTP call), non-JSON response, connection error, JWT claim assertions

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

apps/gateway/src/app.module.ts

@@ -24,6 +24,7 @@ import { GCModule } from './gc/gc.module.js';
 import { ReloadModule } from './reload/reload.module.js';
 import { WorkspaceModule } from './workspace/workspace.module.js';
 import { QueueModule } from './queue/queue.module.js';
+import { FederationModule } from './federation/federation.module.js';
 import { ThrottlerGuard, ThrottlerModule } from '@nestjs/throttler';
 
 @Module({
@@ -52,6 +53,7 @@ import { ThrottlerGuard, ThrottlerModule } from '@nestjs/throttler';
     QueueModule,
     ReloadModule,
     WorkspaceModule,
+    FederationModule,
   ],
   controllers: [HealthController],
   providers: [

apps/gateway/src/federation/ca.dto.ts

@@ -0,0 +1,56 @@
+/**
+ * DTOs for the Step-CA client service (FED-M2-04).
+ *
+ * IssueCertRequestDto  — input to CaService.issueCert()
+ * IssuedCertDto        — output from CaService.issueCert()
+ */
+
+import { IsInt, IsNotEmpty, IsString, IsUUID, Max, Min } from 'class-validator';
+
+export class IssueCertRequestDto {
+  /**
+   * PEM-encoded PKCS#10 Certificate Signing Request.
+   * The CSR must already include the desired SANs.
+   */
+  @IsString()
+  @IsNotEmpty()
+  csrPem!: string;
+
+  /**
+   * UUID of the federation_grants row this certificate is being issued for.
+   * Embedded as the `mosaic_grant_id` custom OID extension.
+   */
+  @IsUUID()
+  grantId!: string;
+
+  /**
+   * UUID of the local user on whose behalf the cert is being issued.
+   * Embedded as the `mosaic_subject_user_id` custom OID extension.
+   */
+  @IsUUID()
+  subjectUserId!: string;
+
+  /**
+   * Requested certificate validity in seconds.
+   * Capped at the step-ca provisioner policy ceiling.
+   * Defaults to 86 400 s (24 h) when omitted by callers.
+   */
+  @IsInt()
+  @Min(60)
+  @Max(365 * 24 * 3600)
+  ttlSeconds!: number;
+}
+
+export class IssuedCertDto {
+  /** PEM-encoded leaf certificate returned by step-ca. */
+  certPem!: string;
+
+  /**
+   * PEM-encoded full certificate chain (leaf + intermediates + root).
+   * Falls back to `certPem` when step-ca returns no `certChain` field.
+   */
+  certChainPem!: string;
+
+  /** Decimal serial number string of the issued certificate. */
+  serialNumber!: string;
+}

apps/gateway/src/federation/ca.service.spec.ts

@@ -0,0 +1,360 @@
+/**
+ * Unit tests for CaService — Step-CA client (FED-M2-04).
+ *
+ * Coverage:
+ *  - Happy path: returns IssuedCertDto with certPem, certChainPem, serialNumber
+ *  - certChainPem fallback: falls back to certPem when certChain absent
+ *  - certChainPem from ca field: uses crt+ca when certChain absent but ca present
+ *  - HTTP 401: throws CaServiceError with cause + remediation
+ *  - HTTP non-401 error: throws CaServiceError
+ *  - Malformed CSR: throws before HTTP call
+ *  - Non-JSON response: throws CaServiceError
+ *  - HTTPS connection error: throws CaServiceError
+ *  - JWT custom claims: mosaic_grant_id and mosaic_subject_user_id present in OTT payload
+ *  - CaServiceError: has cause + remediation properties
+ *  - Missing crt in response: throws CaServiceError
+ */
+
+import { describe, it, expect, vi, beforeEach, type Mock } from 'vitest';
+
+// ---------------------------------------------------------------------------
+// Mock node:https BEFORE importing CaService so the mock is in place when
+// the module is loaded. Vitest/ESM require vi.mock at the top level.
+// ---------------------------------------------------------------------------
+
+vi.mock('node:https', () => {
+  const mockRequest = vi.fn();
+  const mockAgent = vi.fn().mockImplementation(() => ({}));
+  return {
+    default: { request: mockRequest, Agent: mockAgent },
+    request: mockRequest,
+    Agent: mockAgent,
+  };
+});
+
+vi.mock('node:fs', () => {
+  const mockReadFileSync = vi
+    .fn()
+    .mockReturnValue('-----BEGIN CERTIFICATE-----\nFAKEROOT\n-----END CERTIFICATE-----\n');
+  return {
+    default: { readFileSync: mockReadFileSync },
+    readFileSync: mockReadFileSync,
+  };
+});
+
+// ---------------------------------------------------------------------------
+// Helpers
+// ---------------------------------------------------------------------------
+
+// Minimal self-signed certificate PEM produced by openssl for testing.
+// Serial 01, RSA 512 bit (invalid for production, fine for unit tests).
+const FAKE_CERT_PEM = `-----BEGIN CERTIFICATE-----
+MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEA0000000000000000AAAA
+-----END CERTIFICATE-----\n`;
+
+const FAKE_CSR_PEM = `-----BEGIN CERTIFICATE REQUEST-----
+MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEA0000000000000000AAAA
+-----END CERTIFICATE REQUEST-----\n`;
+
+const FAKE_CA_PEM = `-----BEGIN CERTIFICATE-----
+CAROOT000000000000000000000000000000000000000000000000AAAA
+-----END CERTIFICATE-----\n`;
+
+const GRANT_ID = 'a0eebc99-9c0b-4ef8-bb6d-6bb9bd380a11';
+const SUBJECT_USER_ID = 'b1ffcd00-0d1c-5fg9-cc7e-7cc0ce491b22';
+
+// ---------------------------------------------------------------------------
+// Setup env before importing service
+// ---------------------------------------------------------------------------
+
+const JWK_KEY = JSON.stringify({
+  kty: 'oct',
+  kid: 'test-kid',
+  k: 'dGVzdC1zZWNyZXQ=', // base64url("test-secret")
+});
+
+process.env['STEP_CA_URL'] = 'https://step-ca:9000';
+process.env['STEP_CA_PROVISIONER_PASSWORD'] = 'test-password';
+process.env['STEP_CA_PROVISIONER_KEY_JSON'] = JWK_KEY;
+process.env['STEP_CA_ROOT_CERT_PATH'] = '/fake/root.pem';
+
+// Import AFTER env is set and mocks are registered
+import * as httpsModule from 'node:https';
+import { CaService, CaServiceError } from './ca.service.js';
+import type { IssueCertRequestDto } from './ca.dto.js';
+
+// ---------------------------------------------------------------------------
+// Helper to build a mock https.request that simulates step-ca
+// ---------------------------------------------------------------------------
+
+function makeHttpsMock(statusCode: number, body: unknown, errorMsg?: string): void {
+  const mockReq = {
+    write: vi.fn(),
+    end: vi.fn(),
+    on: vi.fn(),
+  };
+
+  (httpsModule.request as unknown as Mock).mockImplementation(
+    (
+      _options: unknown,
+      callback: (res: {
+        statusCode: number;
+        on: (event: string, cb: (chunk?: Buffer) => void) => void;
+      }) => void,
+    ) => {
+      const mockRes = {
+        statusCode,
+        on: (event: string, cb: (chunk?: Buffer) => void) => {
+          if (event === 'data') {
+            if (body !== undefined) {
+              cb(Buffer.from(typeof body === 'string' ? body : JSON.stringify(body)));
+            }
+          }
+          if (event === 'end') {
+            cb();
+          }
+        },
+      };
+
+      if (errorMsg) {
+        // Simulate a connection error via the req.on('error') handler
+        mockReq.on.mockImplementation((event: string, cb: (err: Error) => void) => {
+          if (event === 'error') {
+            setImmediate(() => cb(new Error(errorMsg)));
+          }
+        });
+      } else {
+        // Normal flow: call the response callback
+        setImmediate(() => callback(mockRes));
+      }
+
+      return mockReq;
+    },
+  );
+}
+
+// ---------------------------------------------------------------------------
+// Tests
+// ---------------------------------------------------------------------------
+
+describe('CaService', () => {
+  let service: CaService;
+
+  beforeEach(() => {
+    vi.clearAllMocks();
+    service = new CaService();
+  });
+
+  function makeReq(overrides: Partial<IssueCertRequestDto> = {}): IssueCertRequestDto {
+    return {
+      csrPem: FAKE_CSR_PEM,
+      grantId: GRANT_ID,
+      subjectUserId: SUBJECT_USER_ID,
+      ttlSeconds: 86400,
+      ...overrides,
+    };
+  }
+
+  // -------------------------------------------------------------------------
+  // Happy path
+  // -------------------------------------------------------------------------
+
+  it('returns IssuedCertDto on success (certChain present)', async () => {
+    makeHttpsMock(200, {
+      crt: FAKE_CERT_PEM,
+      certChain: [FAKE_CERT_PEM, FAKE_CA_PEM],
+    });
+
+    const result = await service.issueCert(makeReq());
+
+    expect(result.certPem).toBe(FAKE_CERT_PEM);
+    expect(result.certChainPem).toContain(FAKE_CERT_PEM);
+    expect(result.certChainPem).toContain(FAKE_CA_PEM);
+    expect(typeof result.serialNumber).toBe('string');
+  });
+
+  // -------------------------------------------------------------------------
+  // certChainPem fallback — certChain absent, ca field present
+  // -------------------------------------------------------------------------
+
+  it('builds certChainPem from crt+ca when certChain is absent', async () => {
+    makeHttpsMock(200, {
+      crt: FAKE_CERT_PEM,
+      ca: FAKE_CA_PEM,
+    });
+
+    const result = await service.issueCert(makeReq());
+
+    expect(result.certPem).toBe(FAKE_CERT_PEM);
+    expect(result.certChainPem).toContain(FAKE_CERT_PEM);
+    expect(result.certChainPem).toContain(FAKE_CA_PEM);
+  });
+
+  // -------------------------------------------------------------------------
+  // certChainPem fallback — no certChain, no ca field
+  // -------------------------------------------------------------------------
+
+  it('falls back to certPem alone when certChain and ca are absent', async () => {
+    makeHttpsMock(200, { crt: FAKE_CERT_PEM });
+
+    const result = await service.issueCert(makeReq());
+
+    expect(result.certPem).toBe(FAKE_CERT_PEM);
+    expect(result.certChainPem).toBe(FAKE_CERT_PEM);
+  });
+
+  // -------------------------------------------------------------------------
+  // HTTP 401
+  // -------------------------------------------------------------------------
+
+  it('throws CaServiceError on HTTP 401', async () => {
+    makeHttpsMock(401, { message: 'Unauthorized' });
+
+    await expect(service.issueCert(makeReq())).rejects.toSatisfy((err: unknown) => {
+      if (!(err instanceof CaServiceError)) return false;
+      expect(err.message).toMatch(/401/);
+      expect(err.remediation).toBeTruthy();
+      return true;
+    });
+  });
+
+  // -------------------------------------------------------------------------
+  // HTTP non-401 error (e.g. 422)
+  // -------------------------------------------------------------------------
+
+  it('throws CaServiceError on HTTP 422', async () => {
+    makeHttpsMock(422, { message: 'Unprocessable Entity' });
+
+    await expect(service.issueCert(makeReq())).rejects.toBeInstanceOf(CaServiceError);
+  });
+
+  // -------------------------------------------------------------------------
+  // Malformed CSR — throws before HTTP call
+  // -------------------------------------------------------------------------
+
+  it('throws CaServiceError for malformed CSR without making HTTP call', async () => {
+    const requestSpy = vi.spyOn(httpsModule, 'request');
+
+    await expect(service.issueCert(makeReq({ csrPem: 'not-a-valid-csr' }))).rejects.toBeInstanceOf(
+      CaServiceError,
+    );
+
+    expect(requestSpy).not.toHaveBeenCalled();
+  });
+
+  // -------------------------------------------------------------------------
+  // Non-JSON response
+  // -------------------------------------------------------------------------
+
+  it('throws CaServiceError when step-ca returns non-JSON', async () => {
+    makeHttpsMock(200, 'this is not json');
+
+    await expect(service.issueCert(makeReq())).rejects.toSatisfy((err: unknown) => {
+      if (!(err instanceof CaServiceError)) return false;
+      expect(err.message).toMatch(/non-JSON/);
+      return true;
+    });
+  });
+
+  // -------------------------------------------------------------------------
+  // HTTPS connection error
+  // -------------------------------------------------------------------------
+
+  it('throws CaServiceError on HTTPS connection error', async () => {
+    makeHttpsMock(0, undefined, 'connect ECONNREFUSED 127.0.0.1:9000');
+
+    await expect(service.issueCert(makeReq())).rejects.toSatisfy((err: unknown) => {
+      if (!(err instanceof CaServiceError)) return false;
+      expect(err.message).toMatch(/HTTPS connection/);
+      expect(err.cause).toBeInstanceOf(Error);
+      return true;
+    });
+  });
+
+  // -------------------------------------------------------------------------
+  // JWT custom claims: mosaic_grant_id and mosaic_subject_user_id
+  // -------------------------------------------------------------------------
+
+  it('includes mosaic_grant_id and mosaic_subject_user_id in the OTT payload', async () => {
+    let capturedBody: Record<string, unknown> | undefined;
+
+    // Override the mock to capture the request body
+    const mockReq = {
+      write: vi.fn((data: string) => {
+        capturedBody = JSON.parse(data) as Record<string, unknown>;
+      }),
+      end: vi.fn(),
+      on: vi.fn(),
+    };
+
+    (httpsModule.request as unknown as Mock).mockImplementation(
+      (
+        _options: unknown,
+        callback: (res: {
+          statusCode: number;
+          on: (event: string, cb: (chunk?: Buffer) => void) => void;
+        }) => void,
+      ) => {
+        const mockRes = {
+          statusCode: 200,
+          on: (event: string, cb: (chunk?: Buffer) => void) => {
+            if (event === 'data') {
+              cb(Buffer.from(JSON.stringify({ crt: FAKE_CERT_PEM })));
+            }
+            if (event === 'end') {
+              cb();
+            }
+          },
+        };
+        setImmediate(() => callback(mockRes));
+        return mockReq;
+      },
+    );
+
+    await service.issueCert(makeReq());
+
+    expect(capturedBody).toBeDefined();
+    const ott = capturedBody!['ott'] as string;
+    expect(typeof ott).toBe('string');
+
+    // Decode JWT payload (second segment)
+    const parts = ott.split('.');
+    expect(parts).toHaveLength(3);
+
+    const payloadJson = Buffer.from(parts[1]!, 'base64url').toString('utf8');
+    const payload = JSON.parse(payloadJson) as Record<string, unknown>;
+
+    expect(payload['mosaic_grant_id']).toBe(GRANT_ID);
+    expect(payload['mosaic_subject_user_id']).toBe(SUBJECT_USER_ID);
+  });
+
+  // -------------------------------------------------------------------------
+  // CaServiceError has cause + remediation
+  // -------------------------------------------------------------------------
+
+  it('CaServiceError carries cause and remediation', () => {
+    const cause = new Error('original error');
+    const err = new CaServiceError('something went wrong', 'fix it like this', cause);
+
+    expect(err).toBeInstanceOf(Error);
+    expect(err).toBeInstanceOf(CaServiceError);
+    expect(err.message).toBe('something went wrong');
+    expect(err.remediation).toBe('fix it like this');
+    expect(err.cause).toBe(cause);
+    expect(err.name).toBe('CaServiceError');
+  });
+
+  // -------------------------------------------------------------------------
+  // Missing crt in response
+  // -------------------------------------------------------------------------
+
+  it('throws CaServiceError when response is missing the crt field', async () => {
+    makeHttpsMock(200, { ca: FAKE_CA_PEM });
+
+    await expect(service.issueCert(makeReq())).rejects.toSatisfy((err: unknown) => {
+      if (!(err instanceof CaServiceError)) return false;
+      expect(err.message).toMatch(/missing the "crt" field/);
+      return true;
+    });
+  });
+});

apps/gateway/src/federation/ca.service.ts

@@ -0,0 +1,439 @@
+/**
+ * CaService — Step-CA client for federation grant certificate issuance.
+ *
+ * Responsibilities:
+ *  1. Build a JWK-provisioner One-Time Token (OTT) signed with HS256
+ *     carrying Mosaic-specific claims (`mosaic_grant_id`,
+ *     `mosaic_subject_user_id`, `step.sha`) per the step-ca JWK provisioner
+ *     protocol.
+ *  2. POST the CSR + OTT to the step-ca `/1.0/sign` endpoint over HTTPS,
+ *     pinning the trust to the CA root cert supplied via env.
+ *  3. Return an IssuedCertDto containing the leaf cert, full chain, and
+ *     serial number.
+ *
+ * Environment variables (all required at runtime — validated in constructor):
+ *   STEP_CA_URL                   https://step-ca:9000
+ *   STEP_CA_PROVISIONER_PASSWORD  JWK provisioner password (raw string)
+ *   STEP_CA_PROVISIONER_KEY_JSON  JWK provisioner public+private key (JSON)
+ *   STEP_CA_ROOT_CERT_PATH        Absolute path to the CA root PEM
+ *
+ * Custom OID registry (PRD §6, docs/federation/SETUP.md):
+ *   1.3.6.1.4.1.99999.1  — mosaic_grant_id
+ *   1.3.6.1.4.1.99999.2  — mosaic_subject_user_id
+ *
+ * Fail-loud contract:
+ *   Every error path throws CaServiceError with a human-readable `remediation`
+ *   field. Silent OID-stripping is NEVER allowed — if the sign response does
+ *   not include the cert, we throw rather than return a cert that may be
+ *   missing the custom extensions.
+ */
+
+import { Injectable, Logger } from '@nestjs/common';
+import * as crypto from 'node:crypto';
+import * as fs from 'node:fs';
+import * as https from 'node:https';
+import type { IssueCertRequestDto } from './ca.dto.js';
+import { IssuedCertDto } from './ca.dto.js';
+
+// ---------------------------------------------------------------------------
+// Custom error class
+// ---------------------------------------------------------------------------
+
+export class CaServiceError extends Error {
+  readonly cause: unknown;
+  readonly remediation: string;
+
+  constructor(message: string, remediation: string, cause?: unknown) {
+    super(message);
+    this.name = 'CaServiceError';
+    this.cause = cause;
+    this.remediation = remediation;
+  }
+}
+
+// ---------------------------------------------------------------------------
+// Internal types
+// ---------------------------------------------------------------------------
+
+interface StepSignResponse {
+  crt: string;
+  ca?: string;
+  certChain?: string[];
+}
+
+interface JwkKey {
+  kty: string;
+  kid?: string;
+  use?: string;
+  alg?: string;
+  k?: string; // symmetric
+  n?: string; // RSA
+  e?: string;
+  d?: string;
+  x?: string; // EC
+  y?: string;
+  crv?: string;
+  [key: string]: unknown;
+}
+
+// ---------------------------------------------------------------------------
+// Helpers
+// ---------------------------------------------------------------------------
+
+/**
+ * Base64url-encode a Buffer or string (no padding).
+ */
+function b64url(input: Buffer | string): string {
+  const buf = typeof input === 'string' ? Buffer.from(input, 'utf8') : input;
+  return buf.toString('base64url');
+}
+
+/**
+ * Compute SHA-256 fingerprint of the DER-encoded CSR body.
+ * step-ca uses this as the `step.sha` claim to bind the OTT to a specific CSR.
+ */
+function csrFingerprint(csrPem: string): string {
+  // Strip PEM headers and decode base64 body
+  const b64 = csrPem
+    .replace(/-----BEGIN CERTIFICATE REQUEST-----/, '')
+    .replace(/-----END CERTIFICATE REQUEST-----/, '')
+    .replace(/\s+/g, '');
+
+  let derBuf: Buffer;
+  try {
+    derBuf = Buffer.from(b64, 'base64');
+  } catch (err) {
+    throw new CaServiceError(
+      'Failed to base64-decode the CSR PEM body',
+      'Verify that csrPem is a valid PKCS#10 PEM-encoded certificate request.',
+      err,
+    );
+  }
+
+  if (derBuf.length === 0) {
+    throw new CaServiceError(
+      'CSR PEM decoded to empty buffer — malformed input',
+      'Provide a valid non-empty PKCS#10 PEM-encoded certificate request.',
+    );
+  }
+
+  return crypto.createHash('sha256').update(derBuf).digest('hex');
+}
+
+/**
+ * Derive a signing key from the JWK provisioner password using PBKDF2
+ * then sign with HMAC-SHA256 to produce an HS256 JWT.
+ *
+ * step-ca JWK provisioner tokens:
+ *   - alg: HS256
+ *   - header.kid: provisioner key ID
+ *   - The key is the raw password bytes (step-ca uses the password directly
+ *     as the HMAC key when the JWK provisioner type is "JWK" with symmetric
+ *     key, or the password-derived key when encrypting the JWK).
+ *
+ * Per step-ca source (jose/jwk.go), for a JWK provisioner the OTT is a
+ * JWT signed with the provisioner's decrypted private key.  For HS256 the
+ * key material is the `k` field of the JWK (symmetric secret), which itself
+ * was encrypted with the provisioner password.  Since we already have the
+ * raw provisioner password we use it directly as the HMAC key — this mirrors
+ * what `step ca token` does for symmetric JWK provisioners.
+ */
+function buildOtt(params: {
+  caUrl: string;
+  provisionerPassword: string;
+  provisionerKeyJson: string;
+  csrPem: string;
+  grantId: string;
+  subjectUserId: string;
+  ttlSeconds: number;
+}): string {
+  const {
+    caUrl,
+    provisionerPassword,
+    provisionerKeyJson,
+    csrPem,
+    grantId,
+    subjectUserId,
+    ttlSeconds,
+  } = params;
+
+  let jwk: JwkKey;
+  try {
+    jwk = JSON.parse(provisionerKeyJson) as JwkKey;
+  } catch (err) {
+    throw new CaServiceError(
+      'STEP_CA_PROVISIONER_KEY_JSON is not valid JSON',
+      'Set STEP_CA_PROVISIONER_KEY_JSON to the JSON-serialised JWK object for the mosaic-fed provisioner.',
+      err,
+    );
+  }
+
+  const sha = csrFingerprint(csrPem);
+  const now = Math.floor(Date.now() / 1000);
+  const kid = jwk.kid ?? 'mosaic-fed';
+
+  const header = {
+    alg: 'HS256',
+    typ: 'JWT',
+    kid,
+  };
+
+  const payload = {
+    iss: kid,
+    sub: `${caUrl}/1.0/sign`,
+    aud: [`${caUrl}/1.0/sign`],
+    iat: now,
+    nbf: now - 30, // 30 s clock-skew tolerance
+    exp: now + Math.min(ttlSeconds, 3600), // OTT validity ≤ 1 h
+    sha,
+    // Mosaic custom claims consumed by federation.tpl
+    mosaic_grant_id: grantId,
+    mosaic_subject_user_id: subjectUserId,
+    // step.sha is the canonical field name used in the template
+    step: { sha },
+  };
+
+  const headerB64 = b64url(JSON.stringify(header));
+  const payloadB64 = b64url(JSON.stringify(payload));
+  const signingInput = `${headerB64}.${payloadB64}`;
+
+  // Use the provisioner password as the raw HMAC-SHA256 key.
+  const hmac = crypto.createHmac('sha256', Buffer.from(provisionerPassword, 'utf8'));
+  hmac.update(signingInput);
+  const signature = hmac.digest();
+
+  return `${signingInput}.${b64url(signature)}`;
+}
+
+/**
+ * Send a JSON POST to the step-ca sign endpoint.
+ * Returns the parsed response body or throws CaServiceError.
+ */
+function httpsPost(url: string, body: unknown, agent: https.Agent): Promise<StepSignResponse> {
+  return new Promise((resolve, reject) => {
+    const bodyStr = JSON.stringify(body);
+    const parsed = new URL(url);
+
+    const options: https.RequestOptions = {
+      hostname: parsed.hostname,
+      port: parsed.port ? parseInt(parsed.port, 10) : 443,
+      path: parsed.pathname,
+      method: 'POST',
+      headers: {
+        'Content-Type': 'application/json',
+        'Content-Length': Buffer.byteLength(bodyStr),
+      },
+      agent,
+    };
+
+    const req = https.request(options, (res) => {
+      const chunks: Buffer[] = [];
+      res.on('data', (chunk: Buffer) => chunks.push(chunk));
+      res.on('end', () => {
+        const raw = Buffer.concat(chunks).toString('utf8');
+
+        if (res.statusCode === 401) {
+          reject(
+            new CaServiceError(
+              `step-ca returned HTTP 401 — invalid or expired OTT`,
+              'Check STEP_CA_PROVISIONER_PASSWORD and STEP_CA_PROVISIONER_KEY_JSON. Ensure the mosaic-fed provisioner is configured in the CA.',
+            ),
+          );
+          return;
+        }
+
+        if (res.statusCode && res.statusCode >= 400) {
+          reject(
+            new CaServiceError(
+              `step-ca returned HTTP ${res.statusCode}: ${raw.slice(0, 256)}`,
+              `Review the step-ca logs. Status ${res.statusCode} may indicate a CSR policy violation or misconfigured provisioner.`,
+            ),
+          );
+          return;
+        }
+
+        let parsed: unknown;
+        try {
+          parsed = JSON.parse(raw) as unknown;
+        } catch (err) {
+          reject(
+            new CaServiceError(
+              'step-ca returned a non-JSON response',
+              'Verify STEP_CA_URL points to a running step-ca instance and that TLS is properly configured.',
+              err,
+            ),
+          );
+          return;
+        }
+
+        resolve(parsed as StepSignResponse);
+      });
+    });
+
+    req.on('error', (err: Error) => {
+      reject(
+        new CaServiceError(
+          `HTTPS connection to step-ca failed: ${err.message}`,
+          'Ensure STEP_CA_URL is reachable and STEP_CA_ROOT_CERT_PATH points to the correct CA root certificate.',
+          err,
+        ),
+      );
+    });
+
+    req.write(bodyStr);
+    req.end();
+  });
+}
+
+/**
+ * Extract a decimal serial number from a PEM certificate.
+ * Returns the hex serial if conversion is not possible.
+ */
+function extractSerial(certPem: string): string {
+  try {
+    const cert = new crypto.X509Certificate(certPem);
+    return cert.serialNumber;
+  } catch {
+    return 'unknown';
+  }
+}
+
+// ---------------------------------------------------------------------------
+// Service
+// ---------------------------------------------------------------------------
+
+@Injectable()
+export class CaService {
+  private readonly logger = new Logger(CaService.name);
+
+  private readonly caUrl: string;
+  private readonly provisionerPassword: string;
+  private readonly provisionerKeyJson: string;
+  private readonly rootCertPath: string;
+  private readonly httpsAgent: https.Agent;
+
+  constructor() {
+    const caUrl = process.env['STEP_CA_URL'];
+    const provisionerPassword = process.env['STEP_CA_PROVISIONER_PASSWORD'];
+    const provisionerKeyJson = process.env['STEP_CA_PROVISIONER_KEY_JSON'];
+    const rootCertPath = process.env['STEP_CA_ROOT_CERT_PATH'];
+
+    if (!caUrl) {
+      throw new CaServiceError(
+        'STEP_CA_URL is not set',
+        'Set STEP_CA_URL to the base URL of the step-ca instance, e.g. https://step-ca:9000',
+      );
+    }
+    if (!provisionerPassword) {
+      throw new CaServiceError(
+        'STEP_CA_PROVISIONER_PASSWORD is not set',
+        'Set STEP_CA_PROVISIONER_PASSWORD to the JWK provisioner password for the mosaic-fed provisioner.',
+      );
+    }
+    if (!provisionerKeyJson) {
+      throw new CaServiceError(
+        'STEP_CA_PROVISIONER_KEY_JSON is not set',
+        'Set STEP_CA_PROVISIONER_KEY_JSON to the JSON-encoded JWK for the mosaic-fed provisioner.',
+      );
+    }
+    if (!rootCertPath) {
+      throw new CaServiceError(
+        'STEP_CA_ROOT_CERT_PATH is not set',
+        'Set STEP_CA_ROOT_CERT_PATH to the absolute path of the step-ca root CA certificate PEM file.',
+      );
+    }
+
+    this.caUrl = caUrl;
+    this.provisionerPassword = provisionerPassword;
+    this.provisionerKeyJson = provisionerKeyJson;
+    this.rootCertPath = rootCertPath;
+
+    // Read the root cert and pin it for all HTTPS connections.
+    let rootCert: string;
+    try {
+      rootCert = fs.readFileSync(this.rootCertPath, 'utf8');
+    } catch (err) {
+      throw new CaServiceError(
+        `Cannot read STEP_CA_ROOT_CERT_PATH: ${rootCertPath}`,
+        'Ensure the file exists and is readable by the gateway process.',
+        err,
+      );
+    }
+
+    this.httpsAgent = new https.Agent({
+      ca: rootCert,
+      rejectUnauthorized: true,
+    });
+
+    this.logger.log(`CaService initialised — CA URL: ${this.caUrl}`);
+  }
+
+  /**
+   * Submit a CSR to step-ca and return the issued certificate.
+   *
+   * Throws `CaServiceError` on any failure (network, auth, malformed input).
+   * Never silently swallows errors — fail-loud is a hard contract per M2-02 review.
+   */
+  async issueCert(req: IssueCertRequestDto): Promise<IssuedCertDto> {
+    this.logger.debug(
+      `issueCert — grantId=${req.grantId} subjectUserId=${req.subjectUserId} ttl=${req.ttlSeconds}s`,
+    );
+
+    // Validate CSR before making network calls
+    if (!req.csrPem || !req.csrPem.includes('CERTIFICATE REQUEST')) {
+      throw new CaServiceError(
+        'csrPem does not appear to be a valid PKCS#10 PEM',
+        'Provide a PEM-encoded CSR starting with -----BEGIN CERTIFICATE REQUEST-----.',
+      );
+    }
+
+    const ott = buildOtt({
+      caUrl: this.caUrl,
+      provisionerPassword: this.provisionerPassword,
+      provisionerKeyJson: this.provisionerKeyJson,
+      csrPem: req.csrPem,
+      grantId: req.grantId,
+      subjectUserId: req.subjectUserId,
+      ttlSeconds: req.ttlSeconds,
+    });
+
+    const signUrl = `${this.caUrl}/1.0/sign`;
+    const requestBody = {
+      csr: req.csrPem,
+      ott,
+      validity: {
+        duration: `${req.ttlSeconds}s`,
+      },
+    };
+
+    this.logger.debug(`Posting CSR to ${signUrl}`);
+    const response = await httpsPost(signUrl, requestBody, this.httpsAgent);
+
+    if (!response.crt) {
+      throw new CaServiceError(
+        'step-ca sign response missing the "crt" field',
+        'This is unexpected — the step-ca instance may be misconfigured or running an incompatible version.',
+      );
+    }
+
+    // Build certChainPem: prefer certChain array, fall back to ca field, fall back to crt alone.
+    let certChainPem: string;
+    if (response.certChain && response.certChain.length > 0) {
+      certChainPem = response.certChain.join('\n');
+    } else if (response.ca) {
+      certChainPem = response.crt + '\n' + response.ca;
+    } else {
+      certChainPem = response.crt;
+    }
+
+    const serialNumber = extractSerial(response.crt);
+
+    this.logger.log(`Certificate issued — serial=${serialNumber} grantId=${req.grantId}`);
+
+    const result = new IssuedCertDto();
+    result.certPem = response.crt;
+    result.certChainPem = certChainPem;
+    result.serialNumber = serialNumber;
+    return result;
+  }
+}

apps/gateway/src/federation/federation.module.ts

@@ -0,0 +1,8 @@
+import { Module } from '@nestjs/common';
+import { CaService } from './ca.service.js';
+
+@Module({
+  providers: [CaService],
+  exports: [CaService],
+})
+export class FederationModule {}

deploy/portainer/federated-test.stack.yml

@@ -36,6 +36,12 @@
 #   tested locally — gateway boots, imports resolve, tier-detector runs.
 #   Update digest here when promoting a new build.
 #
+# HEALTHCHECK NOTE (2026-04-21)
+#   Switched from busybox wget to node http.get on 127.0.0.1 (not localhost) to
+#   avoid IPv6 resolution issues on Alpine. Retries increased to 5 and
+#   start_period to 60s to cover the NestJS/GC cold-start window (~40-50s).
+#   restart_policy set to `any` so SIGTERM/clean-exit also triggers restart.
+#
 # NOTE: This is a TEST template — production deployments use a separate
 #       parameterised template with stricter resource limits and secrets.
 
@@ -76,7 +82,7 @@ services:
     deploy:
       replicas: 1
       restart_policy:
-        condition: on-failure
+        condition: any
         delay: 5s
         max_attempts: 3
       labels:
@@ -88,11 +94,15 @@ services:
         - 'traefik.http.routers.${STACK_NAME}.tls.certresolver=letsencrypt'
         - 'traefik.http.services.${STACK_NAME}.loadbalancer.server.port=3000'
     healthcheck:
-      test: ['CMD', 'wget', '-qO-', 'http://localhost:3000/health']
+      test:
+        - 'CMD'
+        - 'node'
+        - '-e'
+        - "require('http').get('http://127.0.0.1:3000/health',r=>process.exit(r.statusCode===200?0:1)).on('error',()=>process.exit(1))"
       interval: 30s
       timeout: 5s
-      retries: 3
-      start_period: 20s
+      retries: 5
+      start_period: 60s
     depends_on:
       - postgres
       - valkey

docs/federation/SETUP.md

@@ -117,3 +117,68 @@ docker compose -f docker-compose.federated.yml logs valkey-federated
 ```
 
 If Valkey is running, verify your firewall allows 6380. On macOS, Docker Desktop may require binding to `host.docker.internal` instead of `localhost`.
+
+## OID Assignments — Mosaic Internal OID Arc
+
+Mosaic uses the private enterprise arc `1.3.6.1.4.1.99999` for custom X.509
+certificate extensions in federation grant certificates.
+
+**IMPORTANT:** This is a development/internal OID arc. Before deploying to a
+production environment accessible by external parties, register a proper IANA
+Private Enterprise Number (PEN) at <https://pen.iana.org/pen/PenApplication.page>
+and update these assignments accordingly.
+
+### Assigned OIDs
+
+| OID                   | Symbolic name                     | Description                                               |
+| --------------------- | --------------------------------- | --------------------------------------------------------- |
+| `1.3.6.1.4.1.99999.1` | `mosaic.federation.grantId`       | UUID of the `federation_grants` row authorising this cert |
+| `1.3.6.1.4.1.99999.2` | `mosaic.federation.subjectUserId` | UUID of the local user on whose behalf the cert is issued |
+
+### Encoding
+
+Each extension value is DER-encoded as an ASN.1 **UTF8String**:
+
+```
+Tag    0x0C        (UTF8String)
+Length 0x24        (36 decimal — fixed length of a UUID string)
+Value  <36 ASCII bytes of the UUID>
+```
+
+The step-ca X.509 template at `infra/step-ca/templates/federation.tpl`
+produces this encoding via the Go template expression:
+
+```
+{{ printf "\x0c\x24%s" .Token.mosaic_grant_id | b64enc }}
+```
+
+The resulting base64 value is passed as the `value` field of the extension
+object in the template JSON.
+
+### CA Environment Variables
+
+The `CaService` (`apps/gateway/src/federation/ca.service.ts`) requires the
+following environment variables at gateway startup:
+
+| Variable                       | Required | Description                                                          |
+| ------------------------------ | -------- | -------------------------------------------------------------------- |
+| `STEP_CA_URL`                  | Yes      | Base URL of the step-ca instance, e.g. `https://step-ca:9000`        |
+| `STEP_CA_PROVISIONER_PASSWORD` | Yes      | JWK provisioner password for the `mosaic-fed` provisioner            |
+| `STEP_CA_PROVISIONER_KEY_JSON` | Yes      | JSON-encoded JWK (public + private) for the `mosaic-fed` provisioner |
+| `STEP_CA_ROOT_CERT_PATH`       | Yes      | Absolute path to the step-ca root CA certificate PEM file            |
+
+Set these variables in your environment or secret manager before starting
+the gateway. In the federated Docker Compose stack they are expected to be
+injected via Docker secrets and environment variable overrides.
+
+### Fail-loud contract
+
+The CA service (and the X.509 template) are designed to fail loudly if the
+custom OIDs cannot be embedded:
+
+- The template produces a malformed extension value (zero-length UTF8String
+  body) when the JWT claims `mosaic_grant_id` or `mosaic_subject_user_id` are
+  absent. step-ca rejects the CSR rather than issuing a cert without the OIDs.
+- `CaService.issueCert()` throws a `CaServiceError` on every error path with
+  a human-readable `remediation` string. It never silently returns a cert that
+  may be missing the required extensions.

infra/step-ca/init.sh

@@ -6,6 +6,10 @@
 # On the first run (no /home/step/config/ca.json present) this script:
 #   1. Initialises Step-CA with a JWK provisioner named "mosaic-fed".
 #   2. Writes the CA configuration to the persistent volume at /home/step.
+#   3. Copies the federation X.509 template into the CA config directory.
+#   4. Patches the mosaic-fed provisioner entry in ca.json to reference the
+#      template via options.x509.templateFile (using jq — must be installed
+#      in the container image).
 #
 # On subsequent runs (config already exists) this script skips init and
 # starts the CA directly.
@@ -18,15 +22,16 @@
 #   Prod: mounted from a Docker secret at /run/secrets/ca_password.
 #
 # OID template:
-#   infra/step-ca/templates/federation.tpl is copied into the CA config
-#   directory so the JWK provisioner can reference it.  The template
-#   skeleton is wired in M2-04 when the CA service lands the SAN-bearing
-#   CSR work.
+#   infra/step-ca/templates/federation.tpl emits custom OID extensions:
+#     1.3.6.1.4.1.99999.1  — mosaic_grant_id
+#     1.3.6.1.4.1.99999.2  — mosaic_subject_user_id
 
 set -e
 
 CA_CONFIG="/home/step/config/ca.json"
 PASSWORD_FILE="/run/secrets/ca_password"
+TEMPLATE_SRC="/etc/step-ca-templates/federation.tpl"
+TEMPLATE_DEST="/home/step/templates/federation.tpl"
 
 if [ ! -f "${CA_CONFIG}" ]; then
   echo "[step-ca init] First boot detected — initialising Mosaic Federation CA..."
@@ -43,12 +48,37 @@ if [ ! -f "${CA_CONFIG}" ]; then
 
   echo "[step-ca init] CA initialised."
 
-  # Copy the X.509 template into the Step-CA config directory so the
-  # provisioner can reference it in M2-04.
-  if [ -f "/etc/step-ca-templates/federation.tpl" ]; then
+  # Copy the X.509 template into the Step-CA config directory.
+  if [ -f "${TEMPLATE_SRC}" ]; then
     mkdir -p /home/step/templates
-    cp /etc/step-ca-templates/federation.tpl /home/step/templates/federation.tpl
-    echo "[step-ca init] Federation X.509 template copied to /home/step/templates/."
+    cp "${TEMPLATE_SRC}" "${TEMPLATE_DEST}"
+    echo "[step-ca init] Federation X.509 template copied to ${TEMPLATE_DEST}."
+  else
+    echo "[step-ca init] WARNING: Template source ${TEMPLATE_SRC} not found — skipping copy."
+  fi
+
+  # Wire the template into the mosaic-fed provisioner via jq.
+  # This is idempotent: the block only runs once (first boot).
+  #
+  # jq filter: find the provisioner entry with name "mosaic-fed" and set
+  # .options.x509.templateFile to the absolute path of the template.
+  # All other provisioners and config keys are left unchanged.
+  if [ -f "${TEMPLATE_DEST}" ] && command -v jq > /dev/null 2>&1; then
+    echo "[step-ca init] Patching mosaic-fed provisioner with X.509 template..."
+    TEMP_CONFIG="${CA_CONFIG}.tmp"
+    jq --arg tpl "${TEMPLATE_DEST}" '
+      .authority.provisioners |= map(
+        if .name == "mosaic-fed" then
+          .options.x509.templateFile = $tpl
+        else
+          .
+        end
+      )
+    ' "${CA_CONFIG}" > "${TEMP_CONFIG}" && mv "${TEMP_CONFIG}" "${CA_CONFIG}"
+    echo "[step-ca init] Provisioner patched."
+  elif ! command -v jq > /dev/null 2>&1; then
+    echo "[step-ca init] WARNING: jq not found — skipping provisioner template patch."
+    echo "[step-ca init] Install jq in the step-ca image to enable automatic template wiring."
   fi
 
   echo "[step-ca init] Startup complete."

infra/step-ca/templates/federation.tpl

@@ -5,41 +5,49 @@
   {{- /*
     Mosaic Federation X.509 Certificate Template
     ============================================
-    This template is used by the "mosaic-fed" JWK provisioner to sign
-    federation client certificates.
+    Provisioner: mosaic-fed (JWK)
+    Implemented: FED-M2-04
 
-    Custom OID extensions (per PRD §6):
-      1.3.6.1.4.1.99999.1  — mosaic.federation.grantId   (UUID string)
-      1.3.6.1.4.1.99999.2  — mosaic.federation.subjectUserId (UUID string)
+    This template emits two custom OID extensions carrying Mosaic federation
+    identifiers. The OTT token (built by CaService.buildOtt) MUST include the
+    claims `mosaic_grant_id` and `mosaic_subject_user_id` as top-level JWT
+    claims. step-ca exposes them under `.Token.<claim>` in this template.
 
-    TODO (M2-04): Wire actual OID extensions below once the CA service
-    (apps/gateway/src/federation/ca.service.ts) lands the SAN-bearing CSR
-    work and the template can be exercised end-to-end.
+    OID Registry (Mosaic Internal Arc — 1.3.6.1.4.1.99999):
+      1.3.6.1.4.1.99999.1  mosaic_grant_id        (UUID, 36 ASCII chars)
+      1.3.6.1.4.1.99999.2  mosaic_subject_user_id  (UUID, 36 ASCII chars)
+
+    DER encoding for each extension value (ASN.1 UTF8String):
+      Tag    = 0x0C  (UTF8String)
+      Length = 0x24  (decimal 36 — the fixed length of a UUID string)
+      Value  = 36 ASCII bytes of the UUID
+
+    The `printf` below builds the raw TLV bytes then base64-encodes them.
+    step-ca expects the `value` field to be base64-encoded raw DER bytes.
+
+    Fail-loud contract:
+      If either claim is missing from the token the printf will produce a
+      zero-length UUID field, making the extension malformed. step-ca will
+      reject the certificate rather than issuing one without the required OIDs.
+      Silent OID stripping is NEVER tolerated.
 
     Step-CA template reference:
       https://smallstep.com/docs/step-ca/templates
-
-    Expected final shape of the extensions block (placeholder — not yet
-    activated):
+  */ -}}
 
   "extensions": [
     {
       "id": "1.3.6.1.4.1.99999.1",
       "critical": false,
-        "value": {{ toJson (first .Token.mosaic_grant_id) }}
+      "value": "{{ printf "\x0c\x24%s" .Token.mosaic_grant_id | b64enc }}"
     },
     {
       "id": "1.3.6.1.4.1.99999.2",
       "critical": false,
-        "value": {{ toJson (first .Token.mosaic_subject_user_id) }}
+      "value": "{{ printf "\x0c\x24%s" .Token.mosaic_subject_user_id | b64enc }}"
     }
   ],
 
-    The provisioner must pass these values in the ACME/JWK token payload
-    (token claims `mosaic_grant_id` and `mosaic_subject_user_id`) when
-    submitting the CSR.  M2-04 owns that work.
-  */ -}}
-
   "keyUsage": ["digitalSignature"],
   "extKeyUsage": ["clientAuth"],
   "basicConstraints": {