Realtime Chat

graph TB
    Client["Client\n(WebSocket/STOMP)"]
    subgraph App["Spring Boot Cluster"]
      App1["App Instance 1\n:8081"]
      App2["App Instance 2\n:8082"]
    end
    subgraph Kafka["Apache Kafka (KRaft)"]
      T1["chat.messages\n(6 partitions, key=roomId)"]
      T2["chat.read-receipts\n(3 partitions)"]
      DLT["DLT (Dead Letter Topics)"]
    end
    subgraph Consumers["Consumer Groups"]
      CG1["chat-persistence\nDB 저장 + 멱등성"]
      CG2["chat-broadcast\nRedis Pub/Sub"]
      CG3["chat-read-receipt\n읽음 처리"]
    end
    Redis["Redis\n(Pub/Sub + Cache)"]
    DB["PostgreSQL"]
    Client -->|STOMP| App1
    Client -->|STOMP| App2
    App1 -->|produce| T1
    App1 -->|produce| T2
    App2 -->|produce| T1
    App2 -->|produce| T2
    T1 -->|consume| CG1
    T1 -->|consume| CG2
    T2 -->|consume| CG3
    CG1 -->|저장| DB
    CG1 -.->|재시도 실패| DLT
    CG2 -->|publish| Redis
    CG3 -->|저장| DB
    Redis -->|subscribe| App1
    Redis -->|subscribe| App2

erDiagram
    users ||--o{ chat_rooms : "created_by"
    users ||--o{ chat_room_members : "참여"
    chat_rooms ||--o{ chat_room_members : "멤버"
    users ||--o{ messages : "발신"
    chat_rooms ||--o{ messages : "소속"

    users {
      bigserial id PK
      varchar email UK
      varchar password
      varchar nickname
      varchar status
      timestamp last_seen_at
      timestamp created_at
    }
    chat_rooms {
      bigserial id PK
      varchar name
      varchar type
      bigint created_by FK
      timestamp created_at
    }
    chat_room_members {
      bigserial id PK
      bigint room_id FK
      bigint user_id FK
      bigint last_read_message_id
      int unread_count
      timestamp joined_at
    }
    messages {
      bigserial id PK
      uuid message_key UK
      bigint room_id FK
      bigint sender_id FK
      text content
      varchar type
      int kafka_partition
      bigint kafka_offset
      timestamp created_at
    }

sequenceDiagram
    participant C as Client
    participant A1 as App-1
    participant K as Kafka
    participant CG1 as chat-persistence
    participant CG2 as chat-broadcast
    participant DB as PostgreSQL
    participant R as Redis
    participant A2 as App-2
    participant C2 as Client (다른 서버)
    C->>A1: STOMP /app/chat.send
    A1->>K: produce (key=roomId)
    par Consumer Group 1
      K->>CG1: consume
      CG1->>DB: 멱등성 체크 + 메시지 저장
      CG1->>DB: unreadCount 증가
    and Consumer Group 2
      K->>CG2: consume
      CG2->>R: PUBLISH room:{roomId}
    end
    R-->>A1: subscribe
    R-->>A2: subscribe
    A1->>C: STOMP /topic/room.{roomId}
    A2->>C2: STOMP /topic/room.{roomId}

sequenceDiagram
    participant C as Client
    participant A as App
    participant K as Kafka
    participant CG3 as chat-read-receipt
    participant DB as PostgreSQL
    participant Cache as Redis Cache
    C->>A: POST /api/rooms/{roomId}/read
    A->>K: produce (chat.read-receipts)
    K->>CG3: consume
    CG3->>DB: unreadCount 초기화
    CG3->>DB: lastReadMessageId 갱신
    CG3->>Cache: evict(userId)

sequenceDiagram
    participant K as Kafka
    participant CG as Consumer
    participant DB as PostgreSQL
    participant DLT as Dead Letter Topic
    K->>CG: consume
    CG->>DB: 처리 시도
    alt 성공
      CG->>K: manual commit
    else 실패 (1~3회 재시도)
      CG->>DB: 재시도
      Note over CG: FixedBackOff 1초 간격, 3회
    else 재시도 실패
      CG->>DLT: DeadLetterPublishingRecoverer
      Note over DLT: 수동 확인 후 재처리
    end