feat(portal-bff): audit log foundation per ADR-0013 #76

Merged
julien merged 1 commits from feat/portal-bff/audit-module into main 2026-05-10 03:44:02 +02:00
9 changed files with 480 additions and 14 deletions
Showing only changes of commit 3a44a1deed - Show all commits
@@ -0,0 +1,67 @@
-- CreateSchema
CREATE SCHEMA IF NOT EXISTS "audit";
-- CreateEnum
CREATE TYPE "audit"."AuditAudience" AS ENUM ('workforce', 'customer');
-- CreateEnum
CREATE TYPE "audit"."AuditOutcome" AS ENUM ('success', 'failure', 'denied');
-- CreateTable
CREATE TABLE "audit"."events" (
"id" UUID NOT NULL,
"created_at" TIMESTAMPTZ(6) NOT NULL DEFAULT CURRENT_TIMESTAMP,
"event_type" TEXT NOT NULL,
"audience" "audit"."AuditAudience" NOT NULL,
"actor_id_hash" TEXT,
"trace_id" TEXT,
"subject" TEXT,
"outcome" "audit"."AuditOutcome" NOT NULL,
"payload" JSONB,
CONSTRAINT "events_pkey" PRIMARY KEY ("id")
);
-- CreateIndex
CREATE INDEX "events_created_at_idx" ON "audit"."events"("created_at");
-- CreateIndex
CREATE INDEX "events_event_type_idx" ON "audit"."events"("event_type");
-- CreateIndex
CREATE INDEX "events_trace_id_idx" ON "audit"."events"("trace_id");
-- ============================================================
-- Append-only contract per ADR-0013
-- ============================================================
-- The schema-level DEFAULT PRIVILEGES set in
-- infra/local/init/postgres/01-init.sql only fire when audit_owner
-- creates the object. This migration runs as a privileged migrator
-- (the same login that owns the public schema), so the table is
-- initially owned by that user and audit_writer/reader/archiver
-- have no grants. Re-apply the owner transfer + grants explicitly
-- so the runtime role contract holds.
ALTER TABLE "audit"."events" OWNER TO audit_owner;
ALTER TYPE "audit"."AuditAudience" OWNER TO audit_owner;
ALTER TYPE "audit"."AuditOutcome" OWNER TO audit_owner;
GRANT INSERT ON "audit"."events" TO audit_writer;
GRANT SELECT ON "audit"."events" TO audit_reader;
-- audit_archiver needs both DELETE *and* SELECT: per ADR-0013 it
-- removes rows "older than retention", which means evaluating a
-- WHERE clause on `created_at`. Postgres requires SELECT on every
-- column referenced in DELETE's WHERE, even when the row count is
-- the only thing returned. SELECT here does NOT widen the contract
-- — UPDATE / TRUNCATE remain ungranted to every role.
GRANT SELECT, DELETE ON "audit"."events" TO audit_archiver;
-- USAGE on the enum types is needed by every role that touches the
-- column. Without it, audit_writer's INSERT fails with "permission
-- denied for type audit.AuditAudience".
GRANT USAGE ON TYPE "audit"."AuditAudience" TO audit_writer, audit_reader, audit_archiver;
GRANT USAGE ON TYPE "audit"."AuditOutcome" TO audit_writer, audit_reader, audit_archiver;
-- No GRANT for UPDATE / TRUNCATE to anyone, including audit_owner
-- at runtime — schema migrations alone amend the table going
-- forward.
@@ -0,0 +1,3 @@
# Please do not edit this file manually
# It should be added in your version-control system (e.g., Git)
provider = "postgresql"
+76 -5
View File
@@ -1,13 +1,84 @@
// This is your Prisma schema file,
// learn more about it in the docs: https://pris.ly/d/prisma-schema
// Get a free hosted Postgres database in seconds: `npx create-db`
// Prisma schema for portal-bff.
//
// `multiSchema` preview is enabled because per ADR-0013 the audit log
// lives in its own `audit` schema with role-based append-only access
// (audit_owner / audit_writer / audit_reader / audit_archiver). The
// public schema holds the regular business data; only audit.events
// lives in audit.
generator client {
provider = "prisma-client-js"
provider = "prisma-client-js"
previewFeatures = ["multiSchema"]
}
datasource db {
provider = "postgresql"
url = env("DATABASE_URL")
schemas = ["public", "audit"]
}
// ============================================================
// Audit log (per ADR-0013)
// ============================================================
//
// Append-only by Postgres role grants — the schema, roles, and
// default privileges are provisioned by infra/local/init/postgres/
// 01-init.sql in dev and the equivalent production manifest. The
// migration that creates this table re-applies the grants
// explicitly and ALTERs the table owner to `audit_owner` so the
// runtime role contract holds even when the migration runs as a
// privileged migrator account.
//
// At runtime, the BFF wraps every INSERT into this table in a
// transaction that begins with `SET LOCAL ROLE audit_writer`, so
// even a compromised BFF connection cannot UPDATE / TRUNCATE /
// DELETE — those grants are not on `audit_writer`.
enum AuditAudience {
workforce
customer
@@schema("audit")
}
enum AuditOutcome {
success
failure
denied
@@schema("audit")
}
model AuditEvent {
id String @id @default(uuid()) @db.Uuid
createdAt DateTime @default(now()) @map("created_at") @db.Timestamptz(6)
eventType String @map("event_type")
audience AuditAudience
// Salted hash (LOG_USER_ID_SALT) of the actor's stable id. NULL
// when the actor is unauthenticated (e.g. failed login attempt
// before resolving an identity). The same salt is used by the
// BFF Pino logger so audit and app logs cross-correlate on this
// field.
actorIdHash String? @map("actor_id_hash")
// W3C trace id (32 hex chars) of the request that produced the
// event. Cross-correlates with traces in Jaeger and with Pino
// log lines that carry the same `trace_id` field. NULL only if
// the event was emitted outside any inbound request (e.g. a
// future cron job).
traceId String? @map("trace_id")
// Free-form identifier of what the event is *about* — typically
// a domain entity URI like `user:42` or `dossier:xyz`. NULL when
// the event is system-wide and has no clear subject.
subject String?
outcome AuditOutcome
// Event-specific structured detail. Redaction of PII is the
// caller's responsibility; the BFF Pino redact list is the
// reference allow-/deny-list.
payload Json?
@@map("events")
@@schema("audit")
@@index([createdAt])
@@index([eventType])
@@index([traceId])
}
+7 -1
View File
@@ -4,9 +4,15 @@ import { AppController } from './app.controller';
import { AppService } from './app.service';
import { ObservabilityModule } from '../observability/observability.module';
import { HealthModule } from '../health/health.module';
import { AuditModule } from '../audit/audit.module';
@Module({
imports: [ObservabilityModule, PrismaModule.forRoot({ isGlobal: true }), HealthModule],
imports: [
ObservabilityModule,
PrismaModule.forRoot({ isGlobal: true }),
AuditModule,
HealthModule,
],
controllers: [AppController],
providers: [AppService],
})
+15
View File
@@ -0,0 +1,15 @@
import { Module } from '@nestjs/common';
import { AuditWriter } from './audit.service';
/**
* Provides the AuditWriter to the rest of the BFF. Imported globally
* by AppModule so any feature module can inject it without an extra
* import. The actual append-only contract is enforced by the
* Postgres role grants set up in the audit-schema migration — see
* apps/portal-bff/prisma/migrations/*_init_audit_schema.
*/
@Module({
providers: [AuditWriter],
exports: [AuditWriter],
})
export class AuditModule {}
@@ -0,0 +1,174 @@
import { Test } from '@nestjs/testing';
import { trace } from '@opentelemetry/api';
import { ClsService } from 'nestjs-cls';
import { PrismaService } from 'nestjs-prisma';
import { Prisma } from '@prisma/client';
import { AuditWriter } from './audit.service';
import type { AuditEventInput } from './audit.types';
interface AuditEventCreateCall {
data: {
eventType: string;
audience: 'workforce' | 'customer';
outcome: 'success' | 'failure' | 'denied';
subject: string | null;
actorIdHash: string | null;
traceId: string | null;
payload: Prisma.InputJsonValue | typeof Prisma.JsonNull;
};
}
interface MockTx {
$executeRawUnsafe: jest.Mock;
auditEvent: { create: jest.Mock };
}
interface MockPrisma {
$transaction: jest.Mock;
tx: MockTx;
}
function buildMocks(): { prisma: MockPrisma; cls: { get: jest.Mock } } {
const tx: MockTx = {
$executeRawUnsafe: jest.fn().mockResolvedValue(0),
auditEvent: { create: jest.fn().mockResolvedValue(undefined) },
};
const prisma: MockPrisma = {
tx,
$transaction: jest.fn(async (fn: (tx: MockTx) => Promise<unknown>) => fn(tx)),
};
const cls = { get: jest.fn() };
return { prisma, cls };
}
async function createSubject(): Promise<{
writer: AuditWriter;
prisma: MockPrisma;
cls: { get: jest.Mock };
}> {
const { prisma, cls } = buildMocks();
const moduleRef = await Test.createTestingModule({
providers: [
AuditWriter,
{ provide: PrismaService, useValue: prisma },
{ provide: ClsService, useValue: cls },
],
}).compile();
return { writer: moduleRef.get(AuditWriter), prisma, cls };
}
const baseInput: AuditEventInput = {
eventType: 'auth.login',
audience: 'workforce',
outcome: 'success',
subject: 'user:42',
};
describe('AuditWriter', () => {
it('locks the transaction to audit_writer before INSERTing', async () => {
const { writer, prisma } = await createSubject();
await writer.recordEvent(baseInput);
expect(prisma.$transaction).toHaveBeenCalledTimes(1);
expect(prisma.tx.$executeRawUnsafe).toHaveBeenCalledWith('SET LOCAL ROLE audit_writer');
// SET ROLE must precede the INSERT, otherwise the runtime
// privilege check happens with the wrong role.
const setRoleOrder = prisma.tx.$executeRawUnsafe.mock.invocationCallOrder[0];
const createOrder = prisma.tx.auditEvent.create.mock.invocationCallOrder[0];
expect(setRoleOrder).toBeLessThan(createOrder);
});
it('passes the input fields through to the create call', async () => {
const { writer, prisma } = await createSubject();
await writer.recordEvent({
...baseInput,
payload: { route: '/auth/login', clientId: 'spa' },
});
const call = prisma.tx.auditEvent.create.mock.calls[0]?.[0] as AuditEventCreateCall;
expect(call.data.eventType).toBe('auth.login');
expect(call.data.audience).toBe('workforce');
expect(call.data.outcome).toBe('success');
expect(call.data.subject).toBe('user:42');
expect(call.data.payload).toEqual({ route: '/auth/login', clientId: 'spa' });
});
it('records Prisma.JsonNull when no payload is provided', async () => {
const { writer, prisma } = await createSubject();
await writer.recordEvent(baseInput);
const call = prisma.tx.auditEvent.create.mock.calls[0]?.[0] as AuditEventCreateCall;
expect(call.data.payload).toBe(Prisma.JsonNull);
});
it('reads actorIdHash from CLS when not passed explicitly', async () => {
const { writer, prisma, cls } = await createSubject();
cls.get.mockReturnValue('hash-from-cls');
await writer.recordEvent(baseInput);
expect(cls.get).toHaveBeenCalledWith('actorIdHash');
const call = prisma.tx.auditEvent.create.mock.calls[0]?.[0] as AuditEventCreateCall;
expect(call.data.actorIdHash).toBe('hash-from-cls');
});
it('prefers an explicit actorIdHash over the CLS-resolved one', async () => {
const { writer, prisma, cls } = await createSubject();
cls.get.mockReturnValue('hash-from-cls');
await writer.recordEvent({ ...baseInput, actorIdHash: 'hash-from-input' });
const call = prisma.tx.auditEvent.create.mock.calls[0]?.[0] as AuditEventCreateCall;
expect(call.data.actorIdHash).toBe('hash-from-input');
});
it('stores actorIdHash = null when neither input nor CLS has one', async () => {
const { writer, prisma, cls } = await createSubject();
cls.get.mockReturnValue(undefined);
await writer.recordEvent(baseInput);
const call = prisma.tx.auditEvent.create.mock.calls[0]?.[0] as AuditEventCreateCall;
expect(call.data.actorIdHash).toBeNull();
});
it('captures the active OTel trace id', async () => {
const fakeSpan = { spanContext: () => ({ traceId: 'abc123', spanId: 'def', traceFlags: 0 }) };
const getActiveSpanSpy = jest.spyOn(trace, 'getActiveSpan').mockReturnValue(fakeSpan as never);
try {
const { writer, prisma } = await createSubject();
await writer.recordEvent(baseInput);
const call = prisma.tx.auditEvent.create.mock.calls[0]?.[0] as AuditEventCreateCall;
expect(call.data.traceId).toBe('abc123');
} finally {
getActiveSpanSpy.mockRestore();
}
});
it('stores traceId = null when no span is active', async () => {
const getActiveSpanSpy = jest.spyOn(trace, 'getActiveSpan').mockReturnValue(undefined);
try {
const { writer, prisma } = await createSubject();
await writer.recordEvent(baseInput);
const call = prisma.tx.auditEvent.create.mock.calls[0]?.[0] as AuditEventCreateCall;
expect(call.data.traceId).toBeNull();
} finally {
getActiveSpanSpy.mockRestore();
}
});
it('propagates the underlying error — no catch-and-swallow', async () => {
const { writer, prisma } = await createSubject();
const dbError = new Error('permission denied for table events');
prisma.tx.auditEvent.create.mockRejectedValueOnce(dbError);
await expect(writer.recordEvent(baseInput)).rejects.toThrow(
'permission denied for table events',
);
});
});
@@ -0,0 +1,76 @@
import { Injectable } from '@nestjs/common';
import { trace } from '@opentelemetry/api';
import { ClsService } from 'nestjs-cls';
import { PrismaService } from 'nestjs-prisma';
import { Prisma } from '@prisma/client';
import type { AuditEventInput } from './audit.types';
/**
* AuditWriter — single entry point for ADR-0013 audit-log writes.
*
* Contract
* --------
* - **Append-only at the database level.** Every write runs inside a
* transaction whose first statement is `SET LOCAL ROLE
* audit_writer`. That role only has `INSERT` on `audit.events`
* (per the migration that created the table); `UPDATE`, `DELETE`,
* `TRUNCATE` all fail at the Postgres level even if the BFF
* connection is otherwise privileged. The role is reset
* automatically at transaction end.
*
* - **Fail loud, never swallow.** Per ADR-0013 §"Blocking writes":
* no audit ⇒ no action. Callers must propagate the rejection up
* so the requested action does not proceed when its audit trail
* cannot be written. The service throws the underlying Prisma
* error unchanged; do not wrap it in a catch-and-log block.
*
* - **trace_id and actor_id_hash are auto-resolved.** trace_id is
* read from the active OTel span context (so the audit row joins
* with the BFF request span and the Pino log lines on the same
* request). actor_id_hash is read from the CLS context populated
* by future auth guards (ADR-0009 / ADR-0010); v1 stores `null`
* when no actor is established. Callers can override either by
* passing them on `AuditEventInput`.
*/
@Injectable()
export class AuditWriter {
constructor(
private readonly prisma: PrismaService,
private readonly cls: ClsService,
) {}
async recordEvent(input: AuditEventInput): Promise<void> {
const traceId = trace.getActiveSpan()?.spanContext().traceId ?? null;
const actorIdHash =
input.actorIdHash ?? this.cls.get<string | undefined>('actorIdHash') ?? null;
await this.prisma.$transaction(async (tx) => {
// Lock the connection to audit_writer for the duration of this
// transaction. SET LOCAL is reset at COMMIT/ROLLBACK so the
// pool's next consumer sees the original role.
await tx.$executeRawUnsafe(`SET LOCAL ROLE audit_writer`);
await tx.auditEvent.create({
data: {
eventType: input.eventType,
audience: input.audience,
outcome: input.outcome,
subject: input.subject ?? null,
actorIdHash,
traceId,
payload: this.toJsonInput(input.payload),
},
});
});
}
// Prisma's `Json` field accepts `Prisma.JsonNull` (null literal in
// SQL JSONB) or a serialisable value; explicit `undefined` skips
// the column. Map `payload` accordingly.
private toJsonInput(
payload: AuditEventInput['payload'],
): Prisma.InputJsonValue | typeof Prisma.JsonNull {
if (payload === undefined) return Prisma.JsonNull;
return payload as Prisma.InputJsonValue;
}
}
+43
View File
@@ -0,0 +1,43 @@
/**
* Public type surface for the audit module — kept narrow so the
* generated Prisma types do not leak through the rest of the BFF.
*/
export type AuditAudience = 'workforce' | 'customer';
export type AuditOutcome = 'success' | 'failure' | 'denied';
export interface AuditEventInput {
/**
* Free-form categorical identifier. Convention: lower-case dotted
* verb-on-noun, e.g. `auth.login`, `dossier.update`. We keep it
* free-form in v1; the catalogue is formalised when we have N
* stable event types in production.
*/
eventType: string;
audience: AuditAudience;
outcome: AuditOutcome;
/**
* What the event is about — typically a domain entity URI like
* `user:42` or `dossier:xyz`. Optional when the event is
* system-wide and has no clear subject.
*/
subject?: string;
/**
* Event-specific structured detail. **Caller-redacted PII** —
* the audit module does not run a redact pass; it is the writer's
* responsibility. Mirror of the Pino redact list in scope.
*/
payload?: Record<string, unknown>;
/**
* Override of the auto-resolved actor id hash. Normally left
* undefined — the service reads from the CLS context (set by
* future auth interceptors per ADR-0009 / ADR-0010). Pass this
* only when emitting events from a context where CLS is not
* populated (background jobs, future cron).
*/
actorIdHash?: string;
}
@@ -156,14 +156,25 @@ Hooks for **admin actions** and **sensitive data access** are designed-in: the w
### Confirmation
- `apps/portal-bff/prisma/schema.prisma` enables `multiSchema` and declares the `audit` schema; the `AuditEvent` model and `AuditOutcome` enum live in it.
- Three Postgres roles exist: `audit_writer` (INSERT only on `audit.events`), `audit_reader` (SELECT only), `audit_archiver` (DELETE only, on rows older than the configured retention). The schema migration includes the explicit `REVOKE UPDATE, DELETE, TRUNCATE FROM PUBLIC` on `audit.events` and the targeted grants.
- `apps/portal-bff/src/audit/audit.module.ts` provides an `AuditService` with one typed method per event family (`signIn`, `signInFailed`, `signOut`, `sessionExpired`, `sessionRevoked`, `tokenValidationFailed`, `mfaAssertionFailed`, `authzDeny`, plus dormant `adminAction` and `sensitiveDataAccess`).
- `AuditService` connects via `AUDIT_DATABASE_URL` as `audit_writer`; a startup probe asserts that the role can `INSERT` and cannot `UPDATE` (a deliberate failing UPDATE during boot is rejected, the BFF starts; if it succeeds, the BFF refuses to start).
- The auth controller, token-validation interceptor, MFA guard, sessions module, and authorization guards each call `AuditService` on the relevant outcomes. Tests assert one audit row per event.
- The retention purge runs daily; failure raises an alert and emits `audit.retention.purge` with `outcome = failure`.
- The same `LOG_USER_ID_SALT` is used by app logs and by audit; an integration test asserts that the same user produces the same `actor_id_hash` in both streams.
- `AUDIT_RETENTION_DAYS < 30` and missing required env vars prevent BFF startup.
**Wired in the foundation PR:**
- `apps/portal-bff/prisma/schema.prisma` enables the `multiSchema` preview, declares the `audit` schema alongside `public`, and carries the `AuditEvent` model with `AuditAudience` (`workforce | customer`) and `AuditOutcome` (`success | failure | denied`) enums.
- The migration `prisma/migrations/*_init_audit_schema/migration.sql` creates `audit.events`, `ALTER`s table + enum types to be owned by `audit_owner`, and re-applies the role grants explicitly: `INSERT` to `audit_writer`, `SELECT` to `audit_reader`, `SELECT, DELETE` to `audit_archiver` (SELECT is needed for archiver to evaluate the `created_at` predicate of "delete older than retention"). No grant of `UPDATE` or `TRUNCATE` to anyone — including the migrator's own login at runtime; only fresh schema migrations amend the table.
- The roles themselves and the schema with default privileges are provisioned earlier by `infra/local/init/postgres/01-init.sql` (dev) — production replicates the same SQL via the future on-prem infrastructure ADR.
- `apps/portal-bff/src/audit/audit.service.ts` exposes a single `AuditWriter.recordEvent(input)` method. Every write runs in a transaction whose first statement is `SET LOCAL ROLE audit_writer`, so the runtime contract holds even if the BFF connection is otherwise privileged. `trace_id` is auto-resolved from the active OTel span; `actor_id_hash` is read from CLS or accepted as an explicit override (placeholder until ADR-0009 / ADR-0010 land their guards). Failures propagate — no catch-and-swallow, per "blocking writes: no audit ⇒ no action".
- BFF connects via the shared `DATABASE_URL` (the role switch is per-transaction). A separate `AUDIT_DATABASE_URL` connection pool is the production hardening, deferred — see "wired as features land" below.
- Smoke-tested end to end against the local-dev Postgres: `audit_writer` INSERTs successfully, fails on `UPDATE` and `DELETE`; `audit_archiver` SELECTs + DELETEs successfully.
**Wired as the corresponding features land:**
- One typed method per event family on `AuditWriter``signIn`, `signInFailed`, `signOut`, `sessionExpired`, `sessionRevoked`, `tokenValidationFailed`, `mfaAssertionFailed`, `authzDeny`, `adminAction`, `sensitiveDataAccess` — added as the matching feature ships (ADR-0009 / ADR-0010 / ADR-0011 / future authz). v1 keeps the surface narrow with the single `recordEvent` so events can be emitted today (e.g. by the auth flow once it lands) while the typed catalogue accretes on a real basis.
- `AUDIT_DATABASE_URL` separate connection (a distinct pool with `audit_writer`-only login credentials) — defense-in-depth that locks down what an at-runtime SQL injection can do further. v1 mitigates with `SET LOCAL ROLE` at the cost of sharing the pool with public-schema reads/writes; v2 splits.
- Startup self-test probe — a deliberate failing `UPDATE` against `audit.events` during boot, asserting it is rejected; if it succeeds, the BFF refuses to start. Lands with the connection split above.
- Retention purge job invoking the `audit_archiver` role daily via cron, emitting `audit.retention.purge` with `outcome = failure` on error. Operational concern — phase-3b infra ADR.
- Auth controller, MFA guard, sessions module, and authorization guards each call `AuditWriter` on the relevant outcomes. Tests assert one audit row per event.
- Integration test verifying that the same user produces the same `actor_id_hash` in both Pino logs and audit rows (same `LOG_USER_ID_SALT`, same hashing path) — wired with the auth + LOG_USER_ID_SALT enforcement.
- `AUDIT_RETENTION_DAYS < 30` and missing required env vars prevent BFF startup — wired with the retention job.
- Live-DB integration tests asserting the role contract (UPDATE/TRUNCATE rejected at runtime) — Testcontainers harness, separate PR.
## Pros and Cons of the Options