aula-04: NGINX Ingress com Keep Request (Lua)

Quando todos os pods estão indisponíveis, o Ingress mantém
a requisição por até 99s aguardando o backend voltar,
alcançando zero falhas visíveis ao usuário.

aula-04/configmap.yaml

@@ -0,0 +1,43 @@
+apiVersion: v1
+kind: ConfigMap
+metadata:
+  name: app-config
+data:
+  MAX_REQUESTS: "3"
+
+  app.js: |
+    const http = require("http");
+
+    const MAX_REQUESTS = Number(process.env.MAX_REQUESTS || 3);
+
+    let requestCount = 0;
+
+    console.log("MAX_REQUESTS =", MAX_REQUESTS);
+
+    const server = http.createServer((req, res) => {
+      if (req.url === "/health") {
+        if (requestCount > MAX_REQUESTS) {
+          console.log('❌ O health check falhou');
+          return;
+        }
+
+        res.writeHead(200);
+        res.end(`ok`);
+        return;
+      }
+
+      requestCount++;
+      console.log("request", requestCount);
+
+      if (requestCount > MAX_REQUESTS) {
+        console.log(`🔥 App travado após ${MAX_REQUESTS} requests`);
+        return;
+      }
+
+      res.writeHead(200);
+      res.end(`Req -> ${requestCount}/${MAX_REQUESTS}`);
+    });
+
+    server.listen(3000, () => {
+      console.log("App rodando na porta 3000");
+    });

aula-04/deployment.yaml

@@ -0,0 +1,47 @@
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: node-bugado
+spec:
+  replicas: 5  # 5 pods - retry encontra pods saudáveis quando alguns falham
+  selector:
+    matchLabels:
+      app: node-bugado
+  template:
+    metadata:
+      labels:
+        app: node-bugado
+    spec:
+      terminationGracePeriodSeconds: 1
+      containers:
+        - name: app
+          image: node:24-alpine
+          command: ["node", "/app/app.js"]
+          env:
+            - name: MAX_REQUESTS
+              valueFrom:
+                configMapKeyRef:
+                  name: app-config
+                  key: MAX_REQUESTS
+          ports:
+            - containerPort: 3000
+          readinessProbe:
+            httpGet:
+              path: /health
+              port: 3000
+            periodSeconds: 1       # Verifica a cada 1 segundo
+            failureThreshold: 1    # 1 falha = remove do Service
+          livenessProbe:
+            httpGet:
+              path: /health
+              port: 3000
+            initialDelaySeconds: 2 # Espera para começar a testar
+            periodSeconds: 1       # A cada 1 seg faz o teste novamente
+            failureThreshold: 2    # 2 falhas = reinicia o container
+          volumeMounts:
+            - name: app-volume
+              mountPath: /app
+      volumes:
+        - name: app-volume
+          configMap:
+            name: app-config

aula-04/ingress-nginx.yaml

@@ -0,0 +1,60 @@
+# NGINX Ingress com "Keep Request" - Mantém request até backend disponível
+# Quando todos os pods estão indisponíveis, aguarda até 99s antes de retornar 503
+apiVersion: networking.k8s.io/v1
+kind: Ingress
+metadata:
+  name: node-bugado-nginx
+  annotations:
+    # Location interno que aguarda o backend ficar disponível
+    nginx.ingress.kubernetes.io/server-snippet: |
+      error_page 502 503 504 = @wait_for_backend;
+
+      location @wait_for_backend {
+        internal;
+        content_by_lua_block {
+          local max_wait = 99
+          local interval = 1
+          local start = ngx.now()
+          local backend_host = "node-bugado.default.svc.cluster.local"
+          local backend_port = 3000
+
+          ngx.log(ngx.INFO, "[KeepRequest] Aguardando backend por ate ", max_wait, "s...")
+
+          while (ngx.now() - start) < max_wait do
+            local sock = ngx.socket.tcp()
+            sock:settimeout(1000)
+
+            local ok, err = sock:connect(backend_host, backend_port)
+            if ok then
+              sock:close()
+              local elapsed = math.floor(ngx.now() - start)
+              ngx.log(ngx.INFO, "[KeepRequest] Backend ready apos ", elapsed, "s")
+              -- Redirect interno (não retorna ao cliente)
+              return ngx.exec(ngx.var.request_uri)
+            end
+
+            ngx.log(ngx.WARN, "[KeepRequest] Backend indisponivel: ", err)
+            ngx.sleep(interval)
+          end
+
+          ngx.status = 503
+          ngx.header["Content-Type"] = "text/plain"
+          ngx.say("Servico indisponivel apos ", max_wait, "s de espera")
+          ngx.exit(503)
+        }
+      }
+    # Timeouts curtos para falhar rápido e ir para o handler Lua
+    nginx.ingress.kubernetes.io/proxy-connect-timeout: "2"
+    nginx.ingress.kubernetes.io/proxy-read-timeout: "3"
+spec:
+  ingressClassName: nginx
+  rules:
+    - http:
+        paths:
+          - path: /
+            pathType: Prefix
+            backend:
+              service:
+                name: node-bugado
+                port:
+                  number: 3000

aula-04/ingress.yaml-

@@ -0,0 +1,26 @@
+# Middleware Retry - tenta outros pods quando um falha
+apiVersion: traefik.io/v1alpha1
+kind: Middleware
+metadata:
+  name: retry-middleware
+spec:
+  retry:
+    attempts: 5           # 5 tentativas
+    initialInterval: 500ms  # 500ms entre ciclos
+---
+# IngressRoute
+apiVersion: traefik.io/v1alpha1
+kind: IngressRoute
+metadata:
+  name: node-bugado
+spec:
+  entryPoints:
+    - web
+  routes:
+    - match: PathPrefix(`/`)
+      kind: Rule
+      middlewares:
+        - name: retry-middleware
+      services:
+        - name: node-bugado
+          port: 3000

aula-04/service.yaml

@@ -0,0 +1,12 @@
+apiVersion: v1
+kind: Service
+metadata:
+  name: node-bugado
+spec:
+  type: LoadBalancer  # Acesso direto na porta 3000 + Traefik na porta 80
+  selector:
+    app: node-bugado
+  ports:
+    - port: 3000
+      targetPort: 3000
+      protocol: TCP

aula-04/teste-stress.sh

@@ -0,0 +1,57 @@
+#!/bin/bash
+
+# Teste de Stress para verificar resiliência do Ingress
+# Uso: ./teste-stress.sh [URL] [NUM_REQUESTS]
+
+URL="${1:-http://localhost}"
+TOTAL="${2:-50}"
+TIMEOUT=120
+
+echo "============================================"
+echo "  Teste de Stress - Alta Disponibilidade"
+echo "============================================"
+echo "URL: $URL"
+echo "Requisições: $TOTAL"
+echo "Timeout: ${TIMEOUT}s"
+echo "============================================"
+echo ""
+
+SUCCESS=0
+FAIL=0
+
+for i in $(seq 1 $TOTAL); do
+    RESULT=$(curl -s -m $TIMEOUT "$URL" 2>&1)
+
+    if echo "$RESULT" | grep -q "Req ->"; then
+        SUCCESS=$((SUCCESS + 1))
+        echo -e "[$i/$TOTAL] \033[32mOK\033[0m - $RESULT"
+    else
+        FAIL=$((FAIL + 1))
+        # Extrai apenas o título do erro se for HTML
+        if echo "$RESULT" | grep -q "<title>"; then
+            ERROR=$(echo "$RESULT" | grep -o "<title>[^<]*</title>" | sed 's/<[^>]*>//g')
+        else
+            ERROR="$RESULT"
+        fi
+        echo -e "[$i/$TOTAL] \033[31mFALHA\033[0m - $ERROR"
+    fi
+
+    sleep 0.2
+done
+
+echo ""
+echo "============================================"
+echo "  Resultado Final"
+echo "============================================"
+PERCENT=$((SUCCESS * 100 / TOTAL))
+echo -e "Sucesso: \033[32m$SUCCESS\033[0m / $TOTAL ($PERCENT%)"
+echo -e "Falhas:  \033[31m$FAIL\033[0m / $TOTAL"
+echo "============================================"
+
+if [ $FAIL -eq 0 ]; then
+    echo -e "\033[32m*** ZERO FALHAS! ***\033[0m"
+elif [ $PERCENT -ge 95 ]; then
+    echo -e "\033[33mBom resultado (>= 95%)\033[0m"
+else
+    echo -e "\033[31mResultado abaixo do esperado\033[0m"
+fi