docs: overhaul deploy.md cleanup and pg verification sections

- add pg connection verification commands (pg_isready, psql queries)
  - renumber pg subsections (4.4.2→4.4.5) to accommodate new section
  - remove MongoDB deploy section (section 4.5) from modelRT deploy guide
  - remove MongoDB SSH tunnel port-forward entries (27017/30017)
  - rewrite section 8 cleanup guide: split into local Docker, local run,
    and K8s(Minikube) categories with scale-down and full-delete options
  - add one-liner kubectl delete -f deploy/k8s/ for full teardown

.gitignore

@@ -22,7 +22,7 @@
 go.work
 
 .vscode
-.idea 
+.idea
 # Shield all log files in the log folder
 /log/
 # Shield config files in the configs folder
@@ -32,6 +32,7 @@ go.work
 # ai config
 .cursor/
 .claude/
+.codewhale/
 .cursorrules
 .copilot/
 .chatgpt/
@@ -39,4 +40,4 @@ go.work
 .vector_cache/
 ai-debug.log
 *.patch
-*.diff
+*.diff

deploy/deploy.md

@@ -695,7 +695,11 @@ kubectl apply -f deploy/k8s/pg-service.yaml
 | **数据库** | `demo` | ConfigMap 中 `POSTGRES_DB` |
 | **用户名** | `postgres` | ConfigMap 中 `POSTGRES_USER` |
 | **密码** | `coslight` | ConfigMap `postgres-config` 中配置，生产环境迁移至 Secret |
-| **存储** | `2Gi` | PVC `postgres-data` |
+| **存储** | `6Gi` | PVC `postgres-data` |
+| **CPU** | `100m` 请求 / `500m` 上限 | StatefulSet `resources` 字段 |
+| **内存** | `256Mi` 请求 / `512Mi` 上限 | StatefulSet `resources` 字段 |
+
+> **注意：** 密码当前以明文形式存储在 `pg-configmap.yaml` 中，生产环境应将其迁移至 K8s Secret，并通过环境变量注入容器，避免将明文密码提交至版本库。
 
 ##### 4.4.1 等待 Pod 就绪
 
@@ -703,7 +707,23 @@ kubectl apply -f deploy/k8s/pg-service.yaml
 kubectl wait --for=condition=ready pod -l app=postgres --timeout=120s
 ```
 
-##### 4.4.2 初始化异步任务表
+##### 4.4.2 连接验证
+
+```bash
+# 快速检查 PostgreSQL 是否接受连接
+kubectl exec -it $(kubectl get pod -l app=postgres -o jsonpath='{.items[0].metadata.name}') \
+  -- pg_isready -U postgres -d demo
+
+# 进入 psql 执行简单查询确认数据库可用
+kubectl exec -it $(kubectl get pod -l app=postgres -o jsonpath='{.items[0].metadata.name}') \
+  -- psql -U postgres -d demo -c "SELECT current_database(), version();"
+
+# 列出所有数据库（确认 demo 库已创建）
+kubectl exec -it $(kubectl get pod -l app=postgres -o jsonpath='{.items[0].metadata.name}') \
+  -- psql -U postgres -c "\l"
+```
+
+##### 4.4.3 初始化异步任务表
 
 PostgreSQL 就绪后执行 1.4 节的建表 SQL，可通过以下方式进入容器执行：
 
@@ -717,14 +737,14 @@ kubectl exec -i $(kubectl get pod -l app=postgres -o jsonpath='{.items[0].metada
   -- psql -U postgres -d demo < /path/to/init.sql
 ```
 
-##### 4.4.3 状态检查
+##### 4.4.4 状态检查
 
 ```bash
 kubectl get pods -l app=postgres
 kubectl logs -l app=postgres --tail=30
 ```
 
-##### 4.4.4 清理
+##### 4.4.5 清理
 
 ```bash
 kubectl delete -f deploy/k8s/pg-service.yaml \
@@ -733,54 +753,6 @@ kubectl delete -f deploy/k8s/pg-service.yaml \
                -f deploy/k8s/pg-configmap.yaml
 ```
 
-#### 4.5 部署 MongoDB
-
-```bash
-kubectl apply -f deploy/k8s/mongodb-secret.yaml
-kubectl apply -f deploy/k8s/mongodb-pvc.yaml
-kubectl apply -f deploy/k8s/mongodb-statefulset.yaml
-kubectl apply -f deploy/k8s/mongodb-service.yaml
-```
-
-| 参数 | 值 | 说明 |
-| :--- | :--- | :--- |
-| **镜像** | `mongo:7.0` | MongoDB 7.0 |
-| **NodePort** | `30017` | 集群外访问端口 |
-| **用户名** | `admin` | Root 管理员 |
-| **密码** | `coslight` | Secret `mongodb-secret` 中配置，生产环境请替换强密码 |
-| **存储** | `2Gi` | PVC `mongodb-data` |
-
-> **注意：** 密码存储在 `mongodb-secret.yaml` 的 `stringData` 中，生产环境应替换为强密码，并避免将明文密码提交至版本库。
-
-##### 4.5.1 等待 Pod 就绪
-
-```bash
-kubectl wait --for=condition=ready pod -l app=mongodb --timeout=120s
-```
-
-##### 4.5.2 连接验证
-
-```bash
-kubectl exec -it $(kubectl get pod -l app=mongodb -o jsonpath='{.items[0].metadata.name}') \
-  -- mongosh -u admin -p coslight --authenticationDatabase admin
-```
-
-##### 4.5.3 状态检查
-
-```bash
-kubectl get pods -l app=mongodb
-kubectl logs -l app=mongodb --tail=30
-```
-
-##### 4.5.4 清理
-
-```bash
-kubectl delete -f deploy/k8s/mongodb-service.yaml \
-               -f deploy/k8s/mongodb-statefulset.yaml \
-               -f deploy/k8s/mongodb-pvc.yaml \
-               -f deploy/k8s/mongodb-secret.yaml
-```
-
 ### 5\. 部署 ModelRT（Kubernetes）
 
 所有资源部署在 `default` 命名空间，YAML 文件位于 `deploy/k8s/`。
@@ -1008,7 +980,6 @@ Mac 本地端口  ──SSH隧道──▶  Ubuntu 宿主机 (192.168.1.101)
 
 ```bash
 ssh -L 5432:192.168.49.2:30432 \
-    -L 27017:192.168.49.2:30017 \
     -L 5671:192.168.49.2:30671 \
     -L 15671:192.168.49.2:31671 \
     -L 6379:192.168.49.2:30001 \
@@ -1024,7 +995,6 @@ ssh -L 5432:192.168.49.2:30432 \
 ```bash
 ssh -fN \
     -L 5432:192.168.49.2:30432 \
-    -L 27017:192.168.49.2:30017 \
     -L 5671:192.168.49.2:30671 \
     -L 15671:192.168.49.2:31671 \
     -L 6379:192.168.49.2:30001 \
@@ -1040,7 +1010,6 @@ ssh -fN \
 | Mac 本地端口 | Minikube NodePort | 服务 | 说明 |
 | :--- | :--- | :--- | :--- |
 | `5432` | `30432` | PostgreSQL | 数据库连接 `localhost:5432` |
-| `27017` | `30017` | MongoDB | 数据库连接 `localhost:27017` |
 | `5671` | `30671` | RabbitMQ AMQP | ModelRT / EventRT 消息队列连接 |
 | `15671` | `31671` | RabbitMQ Management | RabbitMQ 管理界面 `http://localhost:15671` |
 | `6379` | `30001` | Redis | 分布式锁 / 数据存储 |
@@ -1064,14 +1033,111 @@ kill <PID>
 
 ### 8\. 后续操作（停止与清理）
 
-#### 8.1 停止容器
+#### 8.1 本地 Docker 部署清理
+
+适用于第 1、2 节使用 `docker run` 启动的 PostgreSQL 和 Redis 容器。
 
 ```bash
+# 停止容器
 docker stop postgres redis
-```
 
-#### 8.2 删除容器（删除后数据将丢失）
-
-```bash
+# 删除容器（容器内数据将同步丢失）
 docker rm postgres redis
 ```
+
+#### 8.2 本地运行清理
+
+适用于第 3 节以 `go run` 或编译后二进制方式在本地启动的 ModelRT 服务。
+
+前台运行时直接 `Ctrl+C` 终止；后台运行时查找并终止进程：
+
+```bash
+# 终止 go run 启动的进程
+pkill -f "go run main.go"
+
+# 或终止编译后的二进制进程
+pkill model-rt
+```
+
+#### 8.3 K8s(Minikube) 部署清理
+
+适用于第 4、5、6 节在 Minikube 中部署的所有资源。
+
+##### 8.3.1 分服务清理
+
+**仅停止（缩容至 0，PVC 数据保留）**
+
+将所有 Deployment 和 StatefulSet 缩容至 0 副本，Pod 停止运行但持久卷数据不删除，之后可直接缩容回 1 恢复服务。
+
+```bash
+# 停止所有 Deployment（Redis / RabbitMQ / ModelRT / Jaeger / Loki / Grafana）
+kubectl scale deployment --all --replicas=0
+
+# 停止所有 StatefulSet（PostgreSQL，PVC 数据保留）
+kubectl scale statefulset --all --replicas=0
+```
+
+恢复时：
+
+```bash
+kubectl scale deployment --all --replicas=1
+kubectl scale statefulset --all --replicas=1
+```
+
+> **注意：** DaemonSet（Promtail）无法通过 `scale` 停止，如需停用可手动删除其资源：`kubectl delete -f deploy/k8s/promtail-daemonset.yaml`。
+
+---
+
+**永久清理（删除所有资源，包含 PVC，数据不可恢复）**
+
+按部署顺序反向删除各服务资源：
+
+```bash
+# 可观测性栈（Grafana / Promtail / Loki / Jaeger）
+kubectl delete -f deploy/k8s/grafana-service.yaml \
+               -f deploy/k8s/grafana-deployment.yaml \
+               -f deploy/k8s/grafana-configmap.yaml \
+               -f deploy/k8s/promtail-daemonset.yaml \
+               -f deploy/k8s/promtail-configmap.yaml \
+               -f deploy/k8s/promtail-rbac.yaml \
+               -f deploy/k8s/loki-service.yaml \
+               -f deploy/k8s/loki-deployment.yaml \
+               -f deploy/k8s/loki-pvc.yaml \
+               -f deploy/k8s/loki-configmap.yaml \
+               -f deploy/k8s/jaeger-service.yaml \
+               -f deploy/k8s/jaeger-deployment.yaml
+
+# ModelRT 应用
+kubectl delete -f deploy/k8s/modelrt-service.yaml \
+               -f deploy/k8s/modelrt-deployment.yaml \
+               -f deploy/k8s/modelrt-configmap.yaml \
+               -f deploy/k8s/modelrt-secret.yaml
+kubectl delete secret modelrt-certs
+
+# PostgreSQL
+kubectl delete -f deploy/k8s/pg-service.yaml \
+               -f deploy/k8s/pg-statefulset.yaml \
+               -f deploy/k8s/pg-pvc.yaml \
+               -f deploy/k8s/pg-configmap.yaml
+
+# RabbitMQ
+kubectl delete -f deploy/k8s/rabbitmq-service.yaml \
+               -f deploy/k8s/rabbitmq-deployment.yaml \
+               -f deploy/k8s/rabbitmq-users-config.yaml \
+               -f deploy/k8s/rabbitmq-config.yaml \
+               -f deploy/k8s/rabbitmq-secret.yaml
+kubectl delete secret rabbitmq-certs
+
+# Redis
+kubectl delete -f deploy/k8s/redis-service.yaml \
+               -f deploy/k8s/redis-deployment.yaml
+```
+
+##### 8.3.2 一键清理
+
+> **注意：** 此操作会删除 `deploy/k8s/` 下所有 YAML 对应的 K8s 资源，包括 PVC，**持久化数据将永久丢失**，请确认后执行。
+
+```bash
+kubectl delete -f deploy/k8s/
+kubectl delete secret rabbitmq-certs modelrt-certs
+```

deploy/k8s/grafana-deployment.yaml

@@ -16,6 +16,7 @@ spec:
       containers:
         - name: grafana
           image: grafana/grafana:10.4.2
+          imagePullPolicy: IfNotPresent
           ports:
             - containerPort: 3000
           env:

deploy/k8s/jaeger-deployment.yaml

@@ -15,6 +15,7 @@ spec:
       containers:
         - name: jaeger
           image: jaegertracing/all-in-one:1.56
+          imagePullPolicy: IfNotPresent
           env:
             - name: COLLECTOR_OTLP_ENABLED
               value: "true"

deploy/k8s/loki-deployment.yaml

@@ -20,6 +20,7 @@ spec:
       containers:
         - name: loki
           image: grafana/loki:2.9.4
+          imagePullPolicy: IfNotPresent
           args:
             - -config.file=/etc/loki/loki.yaml
           ports:

deploy/k8s/mongodb-statefulset.yaml

@@ -34,9 +34,9 @@ spec:
                 - mongosh
                 - --eval
                 - "db.adminCommand('ping')"
-            initialDelaySeconds: 10
-            periodSeconds: 5
-            timeoutSeconds: 3
+            initialDelaySeconds: 30
+            periodSeconds: 10
+            timeoutSeconds: 10
             failureThreshold: 12
           livenessProbe:
             exec:
@@ -44,10 +44,10 @@ spec:
                 - mongosh
                 - --eval
                 - "db.adminCommand('ping')"
-            initialDelaySeconds: 30
-            periodSeconds: 20
-            timeoutSeconds: 3
-            failureThreshold: 3
+            initialDelaySeconds: 120
+            periodSeconds: 10
+            timeoutSeconds: 30
+            failureThreshold: 5
           resources:
             requests:
               cpu: 100m

deploy/k8s/promtail-daemonset.yaml

@@ -19,6 +19,7 @@ spec:
       containers:
         - name: promtail
           image: grafana/promtail:2.9.4
+          imagePullPolicy: IfNotPresent
           args:
             - -config.file=/etc/promtail/promtail.yaml
           ports:

deploy/k8s/rabbitmq-deployment.yaml

@@ -15,6 +15,7 @@ spec:
       containers:
         - name: rabbitmq
           image: rabbitmq:4.1.1-management-alpine
+          imagePullPolicy: IfNotPresent
           ports:
             - containerPort: 4369
             - containerPort: 5671

deploy/k8s/redis-deployment.yaml

@@ -15,6 +15,7 @@ spec:
       containers:
         - name: redis
           image: redis/redis-stack-server:latest
+          imagePullPolicy: IfNotPresent
           resources:
             limits:
               memory: "128Mi"

diagram/graph.go

@@ -65,9 +65,7 @@ func (g *Graph) AddEdge(from, to uuid.UUID) {
 
 	// 创建新的拓扑信息时，如果被链接的点已经存在于游离节点中
 	// 则将其移除
-	if _, exist := g.FreeVertexs[toKey]; exist {
-		delete(g.FreeVertexs, toKey)
-	}
+	delete(g.FreeVertexs, toKey)
 }
 
 // DelNode delete a node to the graph

logger/loki_syncer.go

@@ -47,8 +47,7 @@ func newLokiSyncer(lCfg config.LokiConfig) *lokiSyncer {
 		client:   &http.Client{Timeout: 5 * time.Second},
 		ch:       make(chan string, 512),
 	}
-	ls.wg.Add(1)
-	go ls.run()
+	ls.wg.Go(ls.run)
 	return ls
 }
 
@@ -70,7 +69,6 @@ func (ls *lokiSyncer) Sync() error {
 }
 
 func (ls *lokiSyncer) run() {
-	defer ls.wg.Done()
 	ticker := time.NewTicker(2 * time.Second)
 	defer ticker.Stop()
 

task/worker.go

@@ -185,13 +185,11 @@ func (w *TaskWorker) Start() error {
 
 	// Start multiple consumers for better throughput
 	for i := 0; i < w.cfg.QueueConsumerCount; i++ {
-		w.wg.Add(1)
-		go w.consumerLoop(i)
+		w.wg.Go(func() { w.consumerLoop(i) })
 	}
 
 	// Start health check goroutine
-	w.wg.Add(1)
-	go w.healthCheckLoop()
+	w.wg.Go(w.healthCheckLoop)
 
 	logger.Info(w.ctx, "task worker started successfully")
 	return nil
@@ -199,8 +197,6 @@ func (w *TaskWorker) Start() error {
 
 // consumerLoop runs a single RabbitMQ consumer
 func (w *TaskWorker) consumerLoop(consumerID int) {
-	defer w.wg.Done()
-
 	logger.Info(w.ctx, "starting consumer", "consumer_id", consumerID)
 
 	// Consume messages from the queue
@@ -478,8 +474,6 @@ func (w *TaskWorker) dispatch(ctx context.Context, taskType TaskType, taskID uui
 
 // healthCheckLoop periodically checks worker health and metrics
 func (w *TaskWorker) healthCheckLoop() {
-	defer w.wg.Done()
-
 	ticker := time.NewTicker(w.cfg.PollingInterval)
 	defer ticker.Stop()