Project Checklist

Toàn bộ hạng mục của dự án ERP/MES — từ planning đến production. Cập nhật theo tiến độ thực tế.

Tiến độ tổng thể theo nhóm

Planning & Docs 100%

Source & CI/CD 80%

Seed Engine 100%

Domain Layer 100%

Application Layer 100%

Infrastructure Layer 100%

API Layer 100%

Testing 100%

Frontend 100%

Deployment 65%

Hoàn thành

Đang làm / Một phần

Chưa làm

Planning & Architecture 100% Hoàn thành

Kiến trúc tổng thể — Composable Architecture docs

ISA-95, CQRS/ES, Event-Driven, Tech Stack, Roadmap 3 phases → erp-mes-architecture.html

Database Design — Event Store + Read Models docs

PostgreSQL DDL, Sensor OEE range-partitioned tables, Redis cache strategy → erp-mes-database.html

Domain Events Catalog — 22 Bounded Contexts docs

Commands, Domain Events, Queries, Sagas cho tất cả 22 BC từ Production đến Sensor OEE → erp-mes-domain-events.html

Industry Seed Engine — 4 ngành done

Jewelry, Cement, Steel, Footwear · ~170 records/ngành · CLI runner → scripts/seed.js

Production BC — DDD Design + C# Code partial

Aggregate, Value Objects, Domain Events, CQRS Handlers → src/Core/Production/

Source Management & CI/CD 80% — Đang hoàn thiện

Platform đã chọn: GitHub + GitHub Actions

GitHub Free (private repos) · GitHub Actions 2,000 phút/tháng miễn phí · GHCR cho container registry

.gitignore, README.md, .env.example done

Cấu hình cơ bản cho .NET 9 + Node.js project

CI Pipeline — GitHub Actions done

.github/workflows/ci.yml · Build, Unit Tests, Architecture Tests (NetArchTest), Integration Tests (Testcontainers), Security Scan (Trivy)

CD Pipeline — Docker Build & Deploy done

.github/workflows/cd.yml · Build Docker image, push GHCR, deploy staging (auto), deploy production (tag + manual approval)

PR Template done

.github/PULL_REQUEST_TEMPLATE.md

Tạo GitHub Repository thật sự

Push code lên GitHub, setup branch protection rules cho main, add collaborators

GitHub Secrets configuration

Add secrets: STAGING_HOST, STAGING_USER, STAGING_SSH_KEY, PROD_HOST, PROD_USER, PROD_SSH_KEY vào GitHub repo settings

GitHub Environments setup

Tạo environments "staging" và "production" với approval rules cho production

Branch protection rules

Require PR review, require CI pass trước khi merge vào main, disable force push

Domain Layer — src/Core/ 100% — 22 Bounded Contexts done ✅

Shared base classes done

AggregateRoot<TId>, DomainEvent, IRepository<T,TId>

Production — ProductionOrder Aggregate done

7 commands, Event Sourcing Apply(), 8 Domain Events, 6 Value Objects, Business Rules enforcement

Quality — QualityInspection Aggregate done

Inspection lifecycle, NCR (NonConformance) sub-entity, verdict logic, InspectionResult value objects

Maintenance — MaintenanceWorkOrder Aggregate done

Work order lifecycle, Asset value object, PM/CM/PdM type distinctions

Inventory / Finance / Auth/IAM / BOM / Planning done

Đầy đủ Aggregate + Value Objects + Domain Events + strongly-typed IDs. Snapshots cho InventoryItem.

Approval / Notification / Traceability / CAPA / Shift / Master Data / Documents done

Procurement / ECO / Calibration / Predictive Analytics / Sensor OEE done

ECO + Calibration ES full lifecycle. Predictive Analytics + Sensor OEE query-side (không ES). OeeFormulas.cs pure helper tại Application layer.

Application Layer — src/Application/ 100% — Tất cả 22 BCs + OeeFormulas + Reporting ✅

CQRS interfaces: ICommand, IQuery, ICommandHandler, IQueryHandler done

Production Commands: Create, Release, Start, RecordOutput, Pause, Resume, Complete, Cancel done

Production Queries: GetById, List done

Quality / Maintenance / Inventory / Finance Commands + Queries done

BOM / Planning / Approval / Notification / Traceability / CAPA / Shift Commands + Queries done

Reporting Engine / OEE / Master Data / Documents / Procurement / ECO / Calibration / Predictive Analytics / Sensor OEE done

OeeFormulas.cs pure static helper (A/P/Q/Total/ShiftWindow) — testable without DB · Reporting query-side với Redis cache

Event Handlers / Projection updaters done

ProductionOrderProjection trong Infrastructure layer — Kafka consumer cập nhật Read Model

Handler wiring — DI (không dùng MediatR) done

Direct DI registration trong ServiceCollectionExtensions.cs. Đơn giản, không phụ thuộc thêm package

Infrastructure Layer — src/Infrastructure/ 100% — 22 BCs + Sensor OEE + Migrations 001-022 ✅

PostgreSQL Event Store Repository done

PostgresProductionOrderRepository: load events → reconstitute, save với optimistic locking, Outbox trong cùng transaction

Outbox Pattern Implementation done

OutboxPublisher BackgroundService: poll 500ms → publish Kafka → UPDATE status. Retry 5 lần trước khi FAILED

Kafka Producer + Consumer setup done

Confluent.Kafka producer (idempotent), consumer group per projection. Headers: event-type cho routing

Production Projection — Kafka → Read Model done

ProductionOrderProjection: xử lý 8 event types, UPDATE production_orders table, idempotent (ON CONFLICT DO NOTHING)

PostgreSQL Read Repository done

PostgresProductionOrderReadRepository: GetById + List với filter status/date. JOIN machine_types + users

Database Migration SQL done

deploy/migrations/001_init_event_store.sql: events, snapshots, outbox, production_orders với indexes đầy đủ

Dependency Injection wiring done

DependencyInjection.cs: register tất cả repositories, producers, background services

Read Model Projections: quality, maintenance done

Read Model Projections: tất cả 15 topics done

Inventory / Finance / BOM / Planning / Approval / Notification / Traceability / CAPA / Shift / Procurement / ECO / Calibration. Tất cả kế thừa KafkaProjectionBase (seek-retry 5× → DLQ projection_failures)

Redis Cache Layer

IDistributedCache (StackExchange.Redis) + IAppCache. Short-TTL: oee:live:{tenant} (5s), reports:ops-dashboard:{tenant} (15s) via CachingReportingReadRepository. Fail-open nếu Redis lỗi. Config: Redis:Cache:*

Database Migrations — SQL thuần (001-022) done

deploy/migrations/: 22 files từ event store → sensor OEE range-partitioned tables. Migration runner trong docker-compose.prod.yml (idempotent psql).

Sensor OEE Data Writer done

Ingest telemetry vào sensor_readings + machine_uptime_events (Postgres range-partitioned); OEE A×P×Q snapshot qua OeeFormulas.cs

API Layer — src/Api/ 100% — 190+ endpoints · SignalR hubs · Rate limiting ✅

ASP.NET Core 9 Minimal API setup done

Program.cs: DI wiring, middleware pipeline, endpoint routing

JWT Authentication + Tenant middleware done

Bearer token validation, tenant_id từ JWT claim hoặc X-Tenant-Id header

Production Order endpoints (10 endpoints) done

Create, Release, Start, RecordOutput, Pause, Resume, Complete, Cancel, GetById, List

Quality endpoints (9) + Maintenance endpoints (9) done

Inspections, NCR flow, work orders, breakdown reporting

Global error handling — Problem Details done

ExceptionHandlingMiddleware: DomainException→422, NotFoundException→404, ArgumentException→400

OpenAPI + Scalar API Explorer done

Scalar UI tại /scalar/v1, JSON spec tại /openapi/v1.json

Health check endpoint done

/health — check PostgreSQL + Redis connectivity, JSON response

Inventory (8) · Finance (8) · Auth (10) · BOM (9) · Planning (10) · Approval (11) · Notification (7) done

Traceability · CAPA · Shift · Reports · OEE · Master Data · Documents · Procurement (12) · ECO (9) · Calibration (7) · Analytics (4) · Sensors done

SignalR real-time hubs done

/hubs/notifications (inbox push) · /hubs/oee (Andon broadcast mỗi 5s, Redis cache 5s TTL)

Bounded Contexts — 22 BCs 22/22 BC done ✅

Production BC done

ProductionOrder aggregate · 8 events · 10 endpoints · 20 unit tests · Event Store + Projection

Quality BC done

QualityInspection + NCR · 7 events · 9 endpoints · 12 unit tests

Maintenance BC done

WorkOrder PM/CM/PdM · AssetBreakdownReported · 7 events · 9 endpoints · 11 unit tests

Inventory BC done

InventoryItem · Receive/Deduct/Reserve/Release/Adjust · LowStockAlert · 8 endpoints · 8 unit tests · stock_levels + stock_movements

Finance BC done

CostEntry Post/Reverse · ProductionCost Calculate · Variance · 8 endpoints · 9 unit tests · cost_entries + production_costs

Auth / IAM done

User state store · Register/Login JWT · RBAC 7 roles · PBKDF2 600k · Refresh token 14d rotation + reuse detection (migration 016) · 10 endpoints · users table

BOM BC done

BillOfMaterials Draft→Active→Obsolete · multi-level explode · ScrapPercent · versioning per product · 9 endpoints · 10 unit tests · bills_of_materials + bom_components

Planning / MRP-lite BC done

ProductionPlan Draft→Released→Closed · RunMrp (BOM explode × Inventory) → Shortage · Dispatch list · 10 endpoints · 10 unit tests

Notification BC · Traceability BC · CAPA BC · Shift Management BC done

Notification: inbox ES + SignalR push · Traceability: lot genealogy forward/backward · CAPA: 7-state lifecycle · Shift: Morning/Afternoon/Night + supervisor + handover

Reporting Engine · OEE Live Andon · Master Data · Documents · Predictive Analytics (query-side) done

Reporting: 6 query types + CSV export + Redis cache · OEE: A×P×Q Andon board SignalR 5s · Master Data: Product/Machine/Employee/Supplier · Documents: SOP/WI/Drawing/Cert lifecycle · Analytics: equipment health + stock shortage + quality risk scores

Procurement BC done

PurchaseOrder ES · Draft→PendingApproval→Approved→(Partial)Received→Closed · Inventory hooks 2 chiều (ReceiveGoods→ReceiveStock; LowStockAlert→auto draft PO) · 12 endpoints · 17 unit tests · migration 018

ECO BC done

EcoRequest Draft→Submitted→Approved(→Implemented auto)|Rejected|Cancelled · ChangeLines AddMaterial/RemoveMaterial/ModifySpec · IEcoBomActivationHook → auto BOM revision · 9 endpoints · migration 020

Calibration BC done

CalibrationRecord Scheduled→InProgress→Passed|Failed|Cancelled · Internal/External/Verification types · nextDueDate > scheduledDate validation · 7 endpoints · migration 021

Sensor OEE (query-side, Postgres range-partitioned) done

sensor_readings + machine_uptime_events + oee_daily_snapshots (partitioned by month, migration 022) · OeeFormulas.cs pure static A/P/Q/Total/ShiftWindow · 35 unit tests · 1 smoke integration test

Approval Workflow Module Matrix + Request + SLA + Delegation + BC hooks ✅

Approval Matrix — cấu hình qua API done

EntityType + RoutingType + Approvers + SlaHours · 1 Active / entity / tenant

ApprovalRequest Aggregate — Event Sourced done

Sequential + Parallel + AnyOf · 8 Domain Events · audit trail · 11 endpoints · 11 unit tests

SLA Engine + Escalation done

DueAt theo SlaHours · EscalateOverdue command khi quá hạn

Delegation có thời hạn done

Ủy quyền 1 lần · ValidUntil · không re-delegate

Integration hooks với BC khác done

IApprovalAutoSubmit: Release ProductionOrder → matrix ProductionOrder; Disposition NCR → NcrDisposition. Thiếu matrix → no-op (BC vẫn succeed).

Tính năng cốt lõi còn thiếu — Phát hiện mới Tất cả done — chỉ còn AI/ML Phase 3

🔴 BOM — Bill of Materials done

Đa cấp · versioning Draft→Active→Obsolete · explode + scrap · 9 endpoints · 10 unit tests · bills_of_materials + bom_components

🔴 Production Planning / MRP-lite done

ProductionPlan · RunMrp (BOM explode + inventory shortage) · Dispatch list · 10 endpoints · 10 unit tests

🔴 Notification System — Centralized done

Inbox ES · SignalR hub · Email/Push stubs · Publish/Read/Dismiss · 7 endpoints · 6 unit tests

🟡 Traceability Engine done

TraceabilityLot Open→Issued→Closed · IssueToOrder · LinkParent (genealogy) · TraceBackward / TraceForward · traceability_lots + lot_links · API /api/v1/traceability/lots

🟡 Reporting Engine done

Daily Production · Quality Summary · Maintenance History · OEE · Ops dashboard · CSV export. API /api/v1/reports + FE /reports.

🟡 OEE Dashboard — Real-time done

Andon board plant/line/machine · SignalR /hubs/oee push mỗi 5s · API /api/v1/oee/live + FE /oee. Sensor/Timescale vẫn future.

🟡 Shift Management done

Schedule · Assign supervisor · Target · Start → Handover → Close. API /api/v1/shifts + FE /shifts.

🟡 CAPA — Corrective & Preventive Action done

Open → Investigate → ActionPlan → InProgress → Verify → Closed. API /api/v1/capa + FE screen.

🟡 Master Data Management done

Product · Machine · Employee · Supplier CRUD (state store). API /api/v1/master-data + FE /master-data.

🟡 Procurement / Purchase Module done 17/07

PurchaseOrder aggregate (ES): Draft → PendingApproval → Approved → PartiallyReceived → Received → Closed · Rejected · Cancelled. Lines nội bộ aggregate, partial receipt per line, chống over-receipt/duplicate material. PO number PO-{year}-{seq} durable. 12 endpoints /api/v1/procurement/purchase-orders · topic procurement.events · migration 018 · 17 unit tests. Inventory hooks 2 chiều (done 18/07, verified live): ① ReceiveGoods → auto ReceiveStock (movement ref PO {số}) · ② LowStockAlert → auto draft PO supplier TBD, qty = max(2×reorder − available, reorder), dedup theo PO mở chứa material. FE /procurement (done 20/07): tạo PO, quản lý lines, submit/approve/reject, nhận hàng từng phần per line, lịch sử receipts, close/cancel

⚪ Document Management

State store controlled_documents (migration 017). Types: SOP/WI/Drawing/MillCert/GIA/NCR. Lifecycle Draft→Published→Obsolete · link entity · version bump. API /api/v1/documents · FE /documents. Blob = storage URI only.

⚪ Engineering Change Order (ECO) done

EcoRequest ES · ECO-{year}-{seq:D5} · ChangeLines dedup · IEcoBomActivationHook auto-activate new BOM revision · 9 endpoints · topic eco.events · migration 020 · FE /eco

⚪ Calibration Management done

CAL-{year}-{seq:D5} · Internal/External/Verification · Scheduled→InProgress→Passed|Failed|Cancelled · nextDueDate validation · certificate_number · 7 endpoints · migration 021 · FE /calibration

⚪ Predictive Analytics — rule-based done

Equipment health score 0-100 (CM frequency + open WOs + overdue cal + low OEE) · Inventory days-to-stockout (Safe/Watch/Warning/Critical) · Quality fail rate per machine (Low/Medium/High/Critical). 4 endpoints /api/v1/analytics · FE /analytics

⚪ AI / ML Anomaly Detection (Phase 3)

Statistical anomaly từ sensor data · Quality defect pattern recognition · Predictive maintenance models. Phụ thuộc vào sensor data đủ lớn (>30 ngày)

Testing 100% — 266 unit · 12 arch · 7 integration ✅

Unit Tests — Domain + Application Layer (xUnit + NSubstitute) done

266 tests: domain business rules tất cả BCs · OeeFormulas 35 tests (A/P/Q/Total/ShiftWindow) · application handler tests (NSubstitute mock repo)

Integration Tests — Infrastructure Layer done

7 smoke tests: Production/Quality/Maintenance/Auth/ECO/Calibration/SensorOEE · Testcontainers Postgres/Redis · assert Event Store + Outbox + DB rows. Project tests/Integration.

Architecture Tests (NetArchTest) — 12 tests done

Layer deps Core→Application→Infrastructure→Api · handlers/events sealed · endpoints namespace. Project tests/Architecture.

API Integration Tests done

WebApplicationFactory · health · register/login · create production order · schedule shift.

Deployment 65% — Docker Compose + CD pipeline + staging server done

Dockerfile (multi-stage, non-root) done

deploy/Dockerfile

Docker Compose — full stack done

deploy/docker-compose.yml: API + PostgreSQL + Redis + Kafka (KRaft)

docker-compose.prod.yml — production overrides

Resource limits (CPU/memory) cho api/worker/postgres/redis/kafka · migration runner service (postgres:17-alpine + psql idempotent) · healthcheck wget · JSON logging (Serilog.Formatting.Json) · ports reset cho infra services

docker-compose.staging.yml

Tương tự prod nhưng ít resource hơn, dùng tag :staging, expose port DB/Redis để debug

VPS Staging server setup done

Nginx reverse proxy · docker-compose.staging.yml · CD SSH deploy tự động từ GitHub Actions khi merge vào staging

Cloudflare Pages — deploy portfolio/docs public CI pipeline ready

.github/workflows/cloudflare-pages.yml — trigger khi docs/ thay đổi trên namdaoint. Cần add 2 GitHub Secrets: CLOUDFLARE_API_TOKEN + CLOUDFLARE_ACCOUNT_ID. Project name: d-pro-portfolio trên Cloudflare Pages dashboard.

HTTPS / SSL setup (Let's Encrypt + Nginx hoặc Traefik)

On-premises deployment guide cho khách hàng nhà máy

Script cài đặt, yêu cầu phần cứng tối thiểu, hướng dẫn IT KH

Database backup strategy

pg_dump scheduled, backup retention, restore procedure

Frontend 100% — Tất cả screens + Operator Tablet done ✅

React + Vite project scaffold done

TypeScript + ESLint tại frontend/ — Vite 5 (Node 18). Proxy /api → localhost:5080

Design System — Custom industrial dark theme done

Navy/Indigo palette · Tabler Icons · Mermaid diagrams pan/zoom + fullscreen. Không dùng MUI/shadcn — custom CSS đủ nhẹ cho shop-floor.

Authentication flow (JWT + Refresh Token) done

apiFetch: 401 → POST /auth/refresh single-flight → retry; fail → clear session + redirect /login. Logout gọi /auth/logout revoke server-side.

Production Order management UI done

List + create orders

Quality / Inventory / Maintenance UI done

BOM / Planning / Finance / Traceability / CAPA / Shifts / Reports / OEE Live / Master Data / Approval / Notifications / Documents / Procurement / ECO / Calibration / Analytics UI done

MES Live Dashboard — real-time machine status, OEE done

Andon board FE /oee + SignalR /hubs/oee (plant / line / machine).

Operator Tablet Mode done

Route /operator không có sidebar — giao diện tối giản cho shop-floor: Record Output + Report Breakdown. Thiết kế touch-friendly.

Security & Performance Hardening — Review 2026-07-14 89% — chỉ còn Kafka HA prod (3 brokers)

PBKDF2 iterations 100k → 600k (OWASP) done

Pbkdf2PasswordHasher.cs — hash cũ vẫn verify được vì iterations lưu trong chuỗi hash

JWT Refresh Token flow done

Access 30 phút + refresh 14 ngày. Rotation mỗi lần refresh; reuse detection → revoke all tokens (REFRESH_TOKEN_REUSED); đổi mật khẩu → revoke all. Bảng refresh_tokens chỉ lưu SHA-256 hash (migration 016). Endpoints: POST /auth/refresh, POST /auth/logout. 7 unit tests

Frontend: auto-refresh access token done

apiFetch: 401 → POST /auth/refresh (single-flight) → retry 1 lần; fail → clear session + redirect /login. Login/logout persist & revoke refresh token.

Snapshot support cho aggregate nhiều events

InventoryItem: IAggregateSnapshotStore + load snapshot + tail events. Write mỗi EventStore:SnapshotEvery (default 100) trong cùng TX với events/outbox. AggregateRoot.Replay / FromSnapshot

Tách Kafka consumers ra Worker process riêng

src/Worker/ host: OutboxPublisher + projections. Flag Workers:RunMessagingInProcess (API default true cho local; compose prod: API=false, Worker=true). OEE SignalR broadcast giữ trên API. Image chung, entrypoint Worker: dotnet ERP.MES.Worker.dll

Kafka production: tối thiểu 3 brokers + replication factor 3

Dev hiện dùng 1 node KRaft — OK cho dev, không đủ HA cho production

Redis: dùng thật hoặc gỡ khỏi stack

Đã dùng Redis cho hot reads (OEE live + ops dashboard). Health check giữ nguyên. Invalidation theo TTL ngắn (shop-floor freshness); projection DEL có thể bổ sung sau.

Rate limiting cho API

.NET AddRateLimiter: global fixed-window per user/tenant+IP (RateLimiting:ApiPermitLimit default 300/60s); policy auth chặt hơn cho login/refresh/register (20/60s). 429 + Retry-After. Exempt: /health, OpenAPI, hubs. Integration tests: RateLimiting:Enabled=false

Number generators → PostgreSQL durable sequence done

Bug cũ: in-memory Interlocked.Increment reset khi restart → UNIQUE violation. Fix: bảng number_sequences (migration 015) + atomic INSERT ON CONFLICT DO UPDATE RETURNING, key per kind/tenant/year — an toàn multi-instance

Operations & Monitoring 80% — chỉ còn Distributed tracing backend

Structured logging (Serilog)

Serilog.AspNetCore + Serilog.Sinks.File cho cả API và Worker. Dev: human-readable console. Production: JSON (Serilog.Formatting.Json) sẵn sàng ingest vào Loki/Cloud Logging. UseSerilogRequestLogging với TenantId enrichment.

Metrics export (OpenTelemetry → Prometheus)

GET /metrics (Prometheus scrape endpoint, exempt khỏi auth + rate limit). ASP.NET Core instrumentation: HTTP request count, duration, status codes. .NET runtime metrics: GC, thread pool, heap size. OTLP export (Tempo/Jaeger) khi cấu hình Telemetry:OtlpEndpoint. Worker có runtime metrics + OTLP.

Grafana dashboards setup

deploy/docker-compose.monitoring.yml: Prometheus (retention 15d) + Grafana :3001, auto-provision datasource + dashboard D-Pro — API & Outbox Overview (request rate, P50/95/99, error rate, outbox pending/published/failures, GC heap, threadpool, in-flight). Custom meter ERP.MES: erp_outbox_pending (gauge ~10s), erp_outbox_published_total, erp_outbox_publish_failures_total

Alerting rules (Prometheus rules)

deploy/monitoring/alerts.yml: ApiDown (1m), HighErrorRate >1% (5m), HighLatencyP99 >500ms (5m), OutboxBacklog >1k (5m), OutboxPublishFailures. Kafka consumer lag: cần kafka-exporter — bổ sung khi có cluster prod

Distributed tracing (OpenTelemetry → Tempo/Jaeger)

🔴 Projection error handling — message bị skip khi lỗi fixed 18/07

Phát hiện khi smoke-test Procurement: event fail → event sau commit offset vượt qua → mất message (cả 13 projections). Fix: KafkaProjectionBase — consume loop chung duy nhất: lỗi → consumer.Seek() retry cùng offset (max 5 lần, delay 2s) → quá 5 lần ghi DLQ projection_failures (payload JSONB + error, migration 019) rồi commit đi tiếp. 13 projections refactor còn mỗi HandleAsync switch (~60 dòng boilerplate/file bỏ đi). BC mới bắt buộc kế thừa base này

Checklist này được cập nhật theo tiến độ thực tế

Cập nhật lần cuối: 22/07/2026

← Về trang chủ