RuoYi Cloud：Gateway、Auth、System 接入 Prometheus JVM 监控，并练习多副本、滚动升级和故障恢复

lee_curry2026-08-06 10:22

RuoYi Cloud：Gateway、Auth、System 接入 Prometheus JVM 监控，并练习多副本、滚动升级和故障恢复

本文基于前面已经部署完成的环境继续操作，不重新安装 MySQL、Redis、Nacos、UI 和监控组件。

本版已修正：

  1. 删除有问题的"批量 scp + ssh + sudo ctr import"命令，改成逐台 Worker 发送、逐台导入、逐台验证。
  2. 删除会报 RuntimeError: 未正确修改 auth.replicas 的跨区块正则脚本，改成按 YAML 顶层服务区块逐行修改并自动校验。
  3. JAR 文件采用自动查找，避免固定文件名与实际构建结果不一致。
  4. ServiceMonitor 标签不再写死，自动读取当前 Prometheus 的选择器标签。

一、当前环境约定

text 复制代码 
命名空间：ruoyi
监控命名空间：monitoring
源码：~/ruoyi-k8s-lab/source/RuoYi-Cloud
Helm Chart：~/ruoyi-k8s-lab/charts/ruoyi-app
镜像构建目录：~/ruoyi-k8s-lab/images
镜像导出目录：~/ruoyi-k8s-lab/artifacts
Worker01：192.168.89.104
Worker02：192.168.89.105
Worker03：192.168.89.106

本阶段只修改：

text 复制代码 
ruoyi-gateway
ruoyi-auth
ruoyi-system

不会重新安装 MySQL、Redis、Nacos、UI、Prometheus、Grafana 和 Alertmanager。

二、部署前检查

在 k8s-master01 执行：

bash 复制代码 
cd "$HOME/ruoyi-k8s-lab"

echo '===== Helm Release ====='
helm list -n ruoyi

echo
echo '===== Deployment ====='
kubectl get deployment -n ruoyi

echo
echo '===== Pod ====='
kubectl get pods -n ruoyi -o wide

echo
echo '===== ServiceMonitor CRD ====='
kubectl get crd servicemonitors.monitoring.coreos.com

echo
echo '===== Prometheus ====='
kubectl get prometheus -n monitoring

必须确认：

text 复制代码 
ruoyi-mysql、ruoyi-redis、ruoyi-nacos、ruoyi-app 均为 deployed
Gateway、Auth、System、UI Pod 均为 Running
ServiceMonitor CRD 存在
Prometheus 资源存在

三、备份当前 Helm Chart

bash 复制代码 
cd "$HOME/ruoyi-k8s-lab"

BACKUP_TIME="$(date +%Y%m%d-%H%M%S)"
cp -a charts/ruoyi-app \
  "charts/ruoyi-app.before-monitoring-${BACKUP_TIME}"

echo "备份完成：charts/ruoyi-app.before-monitoring-${BACKUP_TIME}"

四、确认三个服务已有 Actuator

bash 复制代码 
cd "$HOME/ruoyi-k8s-lab/source/RuoYi-Cloud"

grep -nH \
  'spring-boot-starter-actuator' \
  ruoyi-gateway/pom.xml \
  ruoyi-auth/pom.xml \
  ruoyi-modules/ruoyi-system/pom.xml

三个 POM 都应有输出。

检查 Prometheus Registry 是否已经存在：

bash 复制代码 
grep -nH \
  'micrometer-registry-prometheus' \
  ruoyi-gateway/pom.xml \
  ruoyi-auth/pom.xml \
  ruoyi-modules/ruoyi-system/pom.xml \
  || true

五、添加 Prometheus Registry

bash 复制代码 
cd "$HOME/ruoyi-k8s-lab/source/RuoYi-Cloud"

python3 <<'PY'
from pathlib import Path

pom_files = [
    Path("ruoyi-gateway/pom.xml"),
    Path("ruoyi-auth/pom.xml"),
    Path("ruoyi-modules/ruoyi-system/pom.xml"),
]

dependency = """
        <!-- Micrometer Prometheus 指标注册器 -->
        <dependency>
            <groupId>io.micrometer</groupId>
            <artifactId>micrometer-registry-prometheus</artifactId>
            <scope>runtime</scope>
        </dependency>
"""

for pom in pom_files:
    text = pom.read_text(encoding="utf-8")

    if "micrometer-registry-prometheus" in text:
        print(f"已存在，跳过：{pom}")
        continue

    marker = "    </dependencies>"
    if marker not in text:
        raise RuntimeError(f"未找到 dependencies 结束标签：{pom}")

    text = text.replace(marker, dependency + "\n" + marker, 1)
    pom.write_text(text, encoding="utf-8")
    print(f"添加成功：{pom}")
PY

检查：

bash 复制代码 
grep -n -A 5 -B 2 \
  'micrometer-registry-prometheus' \
  ruoyi-gateway/pom.xml \
  ruoyi-auth/pom.xml \
  ruoyi-modules/ruoyi-system/pom.xml

六、重新编译三个服务

bash 复制代码 
cd "$HOME/ruoyi-k8s-lab/source/RuoYi-Cloud"

export MAVEN_OPTS="-Xms128m -Xmx768m -XX:+UseG1GC"

mvn \
  -DskipTests \
  -T 1 \
  clean package \
  -pl ruoyi-gateway,ruoyi-auth,ruoyi-modules/ruoyi-system \
  -am

成功时最后应出现：

text 复制代码 
BUILD SUCCESS

七、自动定位 JAR 并更新构建目录

bash 复制代码 
cd "$HOME/ruoyi-k8s-lab/source/RuoYi-Cloud"

GATEWAY_JAR="$(
  find ruoyi-gateway/target \
    -maxdepth 1 -type f \
    -name 'ruoyi-gateway*.jar' \
    ! -name '*.original' \
  | sort | head -n 1
)"

AUTH_JAR="$(
  find ruoyi-auth/target \
    -maxdepth 1 -type f \
    -name 'ruoyi-auth*.jar' \
    ! -name '*.original' \
  | sort | head -n 1
)"

SYSTEM_JAR="$(
  find ruoyi-modules/ruoyi-system/target \
    -maxdepth 1 -type f \
    -name 'ruoyi-modules-system*.jar' \
    ! -name '*.original' \
  | sort | head -n 1
)"

for jar in "$GATEWAY_JAR" "$AUTH_JAR" "$SYSTEM_JAR"; do
  if [ -z "$jar" ] || [ ! -s "$jar" ]; then
    echo "JAR 不存在或为空：$jar"
    exit 1
  fi
done

echo "Gateway：$GATEWAY_JAR"
echo "Auth：   $AUTH_JAR"
echo "System： $SYSTEM_JAR"

复制：

bash 复制代码 
mkdir -p \
  "$HOME/ruoyi-k8s-lab/images/gateway" \
  "$HOME/ruoyi-k8s-lab/images/auth" \
  "$HOME/ruoyi-k8s-lab/images/system" \
  "$HOME/ruoyi-k8s-lab/artifacts"

install -m 0644 "$GATEWAY_JAR" \
  "$HOME/ruoyi-k8s-lab/images/gateway/app.jar"

install -m 0644 "$AUTH_JAR" \
  "$HOME/ruoyi-k8s-lab/images/auth/app.jar"

install -m 0644 "$SYSTEM_JAR" \
  "$HOME/ruoyi-k8s-lab/images/system/app.jar"

ls -lh \
  "$HOME/ruoyi-k8s-lab/images/gateway/app.jar" \
  "$HOME/ruoyi-k8s-lab/images/auth/app.jar" \
  "$HOME/ruoyi-k8s-lab/images/system/app.jar"

八、构建 monitoring-v1 镜像

bash 复制代码 
export IMAGE_TAG="3.6.8-monitoring-v1"
bash 复制代码 
sudo buildah bud \
  --format docker --layers \
  -t "localhost/ruoyi-gateway:${IMAGE_TAG}" \
  "$HOME/ruoyi-k8s-lab/images/gateway"
bash 复制代码 
sudo buildah bud \
  --format docker --layers \
  -t "localhost/ruoyi-auth:${IMAGE_TAG}" \
  "$HOME/ruoyi-k8s-lab/images/auth"
bash 复制代码 
sudo buildah bud \
  --format docker --layers \
  -t "localhost/ruoyi-system:${IMAGE_TAG}" \
  "$HOME/ruoyi-k8s-lab/images/system"

检查：

bash 复制代码 
sudo buildah images \
  | grep -E 'ruoyi-(gateway|auth|system)' \
  | grep "$IMAGE_TAG"

九、导出 monitoring-v1 镜像

bash 复制代码 
export IMAGE_TAG="3.6.8-monitoring-v1"
export ARTIFACT_DIR="$HOME/ruoyi-k8s-lab/artifacts"

mkdir -p "$ARTIFACT_DIR"

for service in gateway auth system; do
  rm -f "${ARTIFACT_DIR}/ruoyi-${service}-${IMAGE_TAG}.tar"

  sudo buildah push \
    "localhost/ruoyi-${service}:${IMAGE_TAG}" \
    "docker-archive:${ARTIFACT_DIR}/ruoyi-${service}-${IMAGE_TAG}.tar:localhost/ruoyi-${service}:${IMAGE_TAG}"
done

sudo chown leecurry:leecurry \
  "${ARTIFACT_DIR}/ruoyi-gateway-${IMAGE_TAG}.tar" \
  "${ARTIFACT_DIR}/ruoyi-auth-${IMAGE_TAG}.tar" \
  "${ARTIFACT_DIR}/ruoyi-system-${IMAGE_TAG}.tar"

ls -lh "${ARTIFACT_DIR}"/*"${IMAGE_TAG}".tar

十、修正版：逐台发送并导入 Worker

不再使用原来的批量远程导入命令。必须逐台发送、逐台导入、逐台验证。

10.1 master01 检查文件

bash 复制代码 
export IMAGE_TAG="3.6.8-monitoring-v1"
export ARTIFACT_DIR="$HOME/ruoyi-k8s-lab/artifacts"

for file in \
  "${ARTIFACT_DIR}/ruoyi-gateway-${IMAGE_TAG}.tar" \
  "${ARTIFACT_DIR}/ruoyi-auth-${IMAGE_TAG}.tar" \
  "${ARTIFACT_DIR}/ruoyi-system-${IMAGE_TAG}.tar"
do
  [ -s "$file" ] || {
    echo "镜像文件不存在或为空：$file"
    exit 1
  }
done

10.2 发送到 worker01

bash 复制代码 
scp \
  "${ARTIFACT_DIR}/ruoyi-gateway-${IMAGE_TAG}.tar" \
  "${ARTIFACT_DIR}/ruoyi-auth-${IMAGE_TAG}.tar" \
  "${ARTIFACT_DIR}/ruoyi-system-${IMAGE_TAG}.tar" \
  leecurry@192.168.89.104:/tmp/

10.3 发送到 worker02

bash 复制代码 
scp \
  "${ARTIFACT_DIR}/ruoyi-gateway-${IMAGE_TAG}.tar" \
  "${ARTIFACT_DIR}/ruoyi-auth-${IMAGE_TAG}.tar" \
  "${ARTIFACT_DIR}/ruoyi-system-${IMAGE_TAG}.tar" \
  leecurry@192.168.89.105:/tmp/

10.4 发送到 worker03

bash 复制代码 
scp \
  "${ARTIFACT_DIR}/ruoyi-gateway-${IMAGE_TAG}.tar" \
  "${ARTIFACT_DIR}/ruoyi-auth-${IMAGE_TAG}.tar" \
  "${ARTIFACT_DIR}/ruoyi-system-${IMAGE_TAG}.tar" \
  leecurry@192.168.89.106:/tmp/

10.5 在每台 Worker 分别执行导入

分别切换到 worker01、worker02、worker03 的 Xshell 标签页，每台都执行：

bash 复制代码 
export IMAGE_TAG="3.6.8-monitoring-v1"

for image_file in \
  "/tmp/ruoyi-gateway-${IMAGE_TAG}.tar" \
  "/tmp/ruoyi-auth-${IMAGE_TAG}.tar" \
  "/tmp/ruoyi-system-${IMAGE_TAG}.tar"
do
  [ -s "$image_file" ] || {
    echo "文件不存在或为空：$image_file"
    exit 1
  }

  echo "正在导入：$image_file"
  sudo ctr -n k8s.io images import "$image_file" || exit 1
done

sudo crictl images \
  | grep -E 'ruoyi-(gateway|auth|system)' \
  | grep "$IMAGE_TAG"

每台必须看到三个镜像。三个 Worker 全部验证成功后再继续。

十一、修正版：修改副本数和镜像标签

bash 复制代码 
cd "$HOME/ruoyi-k8s-lab/charts/ruoyi-app"

cp values-lab.yaml \
  "values-lab.yaml.before-monitoring-$(date +%Y%m%d-%H%M%S)"

执行可靠的逐行区块处理脚本：

bash 复制代码 
python3 <<'PY'
from pathlib import Path

path = Path("values-lab.yaml")
lines = path.read_text(encoding="utf-8").splitlines()

services = {"gateway", "auth", "system"}
new_tag = "3.6.8-monitoring-v1"
current_section = None
inside_image = False
replicas_changed = {name: False for name in services}
tags_changed = {name: False for name in services}

for index, line in enumerate(lines):
    stripped = line.strip()

    if line and not line.startswith((" ", "\t")) and stripped.endswith(":"):
        current_section = stripped[:-1]
        inside_image = False
        continue

    if current_section not in services:
        continue

    if line.startswith("  replicas:"):
        lines[index] = "  replicas: 2"
        replicas_changed[current_section] = True
        continue

    if line.startswith("  image:"):
        inside_image = True
        continue

    if inside_image and line.startswith("  ") and not line.startswith("    "):
        inside_image = False

    if inside_image and line.startswith("    tag:"):
        lines[index] = f'    tag: "{new_tag}"'
        tags_changed[current_section] = True

missing_replicas = sorted(
    name for name, changed in replicas_changed.items() if not changed
)
missing_tags = sorted(
    name for name, changed in tags_changed.items() if not changed
)

if missing_replicas:
    raise RuntimeError(
        "以下服务未找到 replicas：" + ", ".join(missing_replicas)
    )

if missing_tags:
    raise RuntimeError(
        "以下服务未找到 image.tag：" + ", ".join(missing_tags)
    )

path.write_text("\n".join(lines) + "\n", encoding="utf-8")

print("values-lab.yaml 修改成功")
for service in ("gateway", "auth", "system"):
    print(f"{service}: replicas=2, tag={new_tag}")
PY

强制校验：

bash 复制代码 
python3 <<'PY'
from pathlib import Path

lines = Path("values-lab.yaml").read_text(
    encoding="utf-8"
).splitlines()

services = {"gateway", "auth", "system"}
expected_tag = "3.6.8-monitoring-v1"
current = None
inside_image = False
result = {
    name: {"replicas": None, "tag": None}
    for name in services
}

for line in lines:
    stripped = line.strip()

    if line and not line.startswith((" ", "\t")) and stripped.endswith(":"):
        current = stripped[:-1]
        inside_image = False
        continue

    if current not in services:
        continue

    if line.startswith("  replicas:"):
        result[current]["replicas"] = stripped.split(":", 1)[1].strip()
        continue

    if line.startswith("  image:"):
        inside_image = True
        continue

    if inside_image and line.startswith("  ") and not line.startswith("    "):
        inside_image = False

    if inside_image and line.startswith("    tag:"):
        result[current]["tag"] = (
            stripped.split(":", 1)[1].strip().strip('"').strip("'")
        )

errors = []
for service in ("gateway", "auth", "system"):
    if result[service]["replicas"] != "2":
        errors.append(f"{service}.replicas={result[service]['replicas']}")

    if result[service]["tag"] != expected_tag:
        errors.append(f"{service}.image.tag={result[service]['tag']}")

if errors:
    raise SystemExit("校验失败：\n- " + "\n- ".join(errors))

print("校验通过：")
for service in ("gateway", "auth", "system"):
    print(f"- {service}: replicas=2, tag={expected_tag}")
PY

查看：

bash 复制代码 
grep -nE \
  '^(gateway|auth|system|ui):|^  replicas:|^  image:|^    repository:|^    tag:' \
  values-lab.yaml

Gateway/Auth/System 应为 replicas: 2 和 monitoring-v1；UI 保持原值。

十二、更新 Gateway Deployment

bash 复制代码 
cd "$HOME/ruoyi-k8s-lab/charts/ruoyi-app"

cat <<'EOF' > templates/gateway-deployment.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
  name: {{ .Values.gateway.name }}
  labels:
    app.kubernetes.io/name: {{ .Values.gateway.name }}
    app.kubernetes.io/instance: {{ .Release.Name }}
spec:
  replicas: {{ .Values.gateway.replicas }}
  revisionHistoryLimit: 5

  strategy:
    type: RollingUpdate
    rollingUpdate:
      maxUnavailable: 1
      maxSurge: 0

  selector:
    matchLabels:
      app.kubernetes.io/name: {{ .Values.gateway.name }}
      app.kubernetes.io/instance: {{ .Release.Name }}

  template:
    metadata:
      labels:
        app.kubernetes.io/name: {{ .Values.gateway.name }}
        app.kubernetes.io/instance: {{ .Release.Name }}

    spec:
      affinity:
        podAntiAffinity:
          preferredDuringSchedulingIgnoredDuringExecution:
            - weight: 100
              podAffinityTerm:
                topologyKey: kubernetes.io/hostname
                labelSelector:
                  matchLabels:
                    app.kubernetes.io/name: {{ .Values.gateway.name }}
                    app.kubernetes.io/instance: {{ .Release.Name }}

      initContainers:
        - name: wait-for-nacos
          image: "{{ .Values.global.busyboxImage.repository }}:{{ .Values.global.busyboxImage.tag }}"
          imagePullPolicy: IfNotPresent
          command:
            - sh
            - -c
            - |
              until nc -z ruoyi-nacos 8848; do
                echo "等待 Nacos"
                sleep 3
              done

        - name: wait-for-redis
          image: "{{ .Values.global.busyboxImage.repository }}:{{ .Values.global.busyboxImage.tag }}"
          imagePullPolicy: IfNotPresent
          command:
            - sh
            - -c
            - |
              until nc -z ruoyi-redis 6379; do
                echo "等待 Redis"
                sleep 3
              done

      containers:
        - name: gateway
          image: "{{ .Values.gateway.image.repository }}:{{ .Values.gateway.image.tag }}"
          imagePullPolicy: {{ .Values.global.imagePullPolicy }}

          env:
            - name: JAVA_OPTS
              value: {{ .Values.gateway.javaOpts | quote }}

            - name: SPRING_PROFILES_ACTIVE
              value: {{ .Values.global.profile | quote }}

            - name: MANAGEMENT_ENDPOINTS_WEB_EXPOSURE_INCLUDE
              value: "health,info,prometheus"

            - name: MANAGEMENT_ENDPOINT_PROMETHEUS_ENABLED
              value: "true"

            - name: MANAGEMENT_PROMETHEUS_METRICS_EXPORT_ENABLED
              value: "true"

            - name: MANAGEMENT_METRICS_TAGS_APPLICATION
              value: {{ .Values.gateway.name | quote }}

            - name: SPRING_CLOUD_NACOS_SERVER_ADDR
              value: {{ .Values.global.nacos.serverAddr | quote }}

            - name: SPRING_CLOUD_NACOS_CONFIG_SERVER_ADDR
              value: {{ .Values.global.nacos.serverAddr | quote }}

            - name: SPRING_CLOUD_NACOS_DISCOVERY_SERVER_ADDR
              value: {{ .Values.global.nacos.serverAddr | quote }}

            - name: SPRING_CLOUD_NACOS_USERNAME
              valueFrom:
                secretKeyRef:
                  name: ruoyi-app-nacos
                  key: username

            - name: SPRING_CLOUD_NACOS_PASSWORD
              valueFrom:
                secretKeyRef:
                  name: ruoyi-app-nacos
                  key: password

            - name: SPRING_CLOUD_NACOS_CONFIG_USERNAME
              valueFrom:
                secretKeyRef:
                  name: ruoyi-app-nacos
                  key: username

            - name: SPRING_CLOUD_NACOS_CONFIG_PASSWORD
              valueFrom:
                secretKeyRef:
                  name: ruoyi-app-nacos
                  key: password

            - name: SPRING_CLOUD_NACOS_DISCOVERY_USERNAME
              valueFrom:
                secretKeyRef:
                  name: ruoyi-app-nacos
                  key: username

            - name: SPRING_CLOUD_NACOS_DISCOVERY_PASSWORD
              valueFrom:
                secretKeyRef:
                  name: ruoyi-app-nacos
                  key: password

          ports:
            - name: http
              containerPort: {{ .Values.gateway.port }}

          startupProbe:
            tcpSocket:
              port: http
            periodSeconds: 5
            failureThreshold: 60

          readinessProbe:
            tcpSocket:
              port: http
            periodSeconds: 10
            failureThreshold: 6

          livenessProbe:
            tcpSocket:
              port: http
            periodSeconds: 20
            failureThreshold: 6

          resources:
{{ toYaml .Values.gateway.resources | indent 12 }}
EOF

十三、更新 Auth Deployment

bash 复制代码 
cd "$HOME/ruoyi-k8s-lab/charts/ruoyi-app"

cat <<'EOF' > templates/auth-deployment.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
  name: {{ .Values.auth.name }}
  labels:
    app.kubernetes.io/name: {{ .Values.auth.name }}
    app.kubernetes.io/instance: {{ .Release.Name }}
spec:
  replicas: {{ .Values.auth.replicas }}
  revisionHistoryLimit: 5

  strategy:
    type: RollingUpdate
    rollingUpdate:
      maxUnavailable: 1
      maxSurge: 0

  selector:
    matchLabels:
      app.kubernetes.io/name: {{ .Values.auth.name }}
      app.kubernetes.io/instance: {{ .Release.Name }}

  template:
    metadata:
      labels:
        app.kubernetes.io/name: {{ .Values.auth.name }}
        app.kubernetes.io/instance: {{ .Release.Name }}

    spec:
      affinity:
        podAntiAffinity:
          preferredDuringSchedulingIgnoredDuringExecution:
            - weight: 100
              podAffinityTerm:
                topologyKey: kubernetes.io/hostname
                labelSelector:
                  matchLabels:
                    app.kubernetes.io/name: {{ .Values.auth.name }}
                    app.kubernetes.io/instance: {{ .Release.Name }}

      initContainers:
        - name: wait-for-nacos
          image: "{{ .Values.global.busyboxImage.repository }}:{{ .Values.global.busyboxImage.tag }}"
          imagePullPolicy: IfNotPresent
          command:
            - sh
            - -c
            - |
              until nc -z ruoyi-nacos 8848; do
                echo "等待 Nacos"
                sleep 3
              done

        - name: wait-for-redis
          image: "{{ .Values.global.busyboxImage.repository }}:{{ .Values.global.busyboxImage.tag }}"
          imagePullPolicy: IfNotPresent
          command:
            - sh
            - -c
            - |
              until nc -z ruoyi-redis 6379; do
                echo "等待 Redis"
                sleep 3
              done

      containers:
        - name: auth
          image: "{{ .Values.auth.image.repository }}:{{ .Values.auth.image.tag }}"
          imagePullPolicy: {{ .Values.global.imagePullPolicy }}

          env:
            - name: JAVA_OPTS
              value: {{ .Values.auth.javaOpts | quote }}

            - name: SPRING_PROFILES_ACTIVE
              value: {{ .Values.global.profile | quote }}

            - name: MANAGEMENT_ENDPOINTS_WEB_EXPOSURE_INCLUDE
              value: "health,info,prometheus"

            - name: MANAGEMENT_ENDPOINT_PROMETHEUS_ENABLED
              value: "true"

            - name: MANAGEMENT_PROMETHEUS_METRICS_EXPORT_ENABLED
              value: "true"

            - name: MANAGEMENT_METRICS_TAGS_APPLICATION
              value: {{ .Values.auth.name | quote }}

            - name: SPRING_CLOUD_NACOS_SERVER_ADDR
              value: {{ .Values.global.nacos.serverAddr | quote }}

            - name: SPRING_CLOUD_NACOS_CONFIG_SERVER_ADDR
              value: {{ .Values.global.nacos.serverAddr | quote }}

            - name: SPRING_CLOUD_NACOS_DISCOVERY_SERVER_ADDR
              value: {{ .Values.global.nacos.serverAddr | quote }}

            - name: SPRING_CLOUD_NACOS_USERNAME
              valueFrom:
                secretKeyRef:
                  name: ruoyi-app-nacos
                  key: username

            - name: SPRING_CLOUD_NACOS_PASSWORD
              valueFrom:
                secretKeyRef:
                  name: ruoyi-app-nacos
                  key: password

            - name: SPRING_CLOUD_NACOS_CONFIG_USERNAME
              valueFrom:
                secretKeyRef:
                  name: ruoyi-app-nacos
                  key: username

            - name: SPRING_CLOUD_NACOS_CONFIG_PASSWORD
              valueFrom:
                secretKeyRef:
                  name: ruoyi-app-nacos
                  key: password

            - name: SPRING_CLOUD_NACOS_DISCOVERY_USERNAME
              valueFrom:
                secretKeyRef:
                  name: ruoyi-app-nacos
                  key: username

            - name: SPRING_CLOUD_NACOS_DISCOVERY_PASSWORD
              valueFrom:
                secretKeyRef:
                  name: ruoyi-app-nacos
                  key: password

          ports:
            - name: http
              containerPort: {{ .Values.auth.port }}

          startupProbe:
            tcpSocket:
              port: http
            periodSeconds: 5
            failureThreshold: 60

          readinessProbe:
            tcpSocket:
              port: http
            periodSeconds: 10
            failureThreshold: 6

          livenessProbe:
            tcpSocket:
              port: http
            periodSeconds: 20
            failureThreshold: 6

          resources:
{{ toYaml .Values.auth.resources | indent 12 }}
EOF

十四、更新 System Deployment

bash 复制代码 
cd "$HOME/ruoyi-k8s-lab/charts/ruoyi-app"

cat <<'EOF' > templates/system-deployment.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
  name: {{ .Values.system.name }}
  labels:
    app.kubernetes.io/name: {{ .Values.system.name }}
    app.kubernetes.io/instance: {{ .Release.Name }}
spec:
  replicas: {{ .Values.system.replicas }}
  revisionHistoryLimit: 5

  strategy:
    type: RollingUpdate
    rollingUpdate:
      maxUnavailable: 1
      maxSurge: 0

  selector:
    matchLabels:
      app.kubernetes.io/name: {{ .Values.system.name }}
      app.kubernetes.io/instance: {{ .Release.Name }}

  template:
    metadata:
      labels:
        app.kubernetes.io/name: {{ .Values.system.name }}
        app.kubernetes.io/instance: {{ .Release.Name }}

    spec:
      affinity:
        podAntiAffinity:
          preferredDuringSchedulingIgnoredDuringExecution:
            - weight: 100
              podAffinityTerm:
                topologyKey: kubernetes.io/hostname
                labelSelector:
                  matchLabels:
                    app.kubernetes.io/name: {{ .Values.system.name }}
                    app.kubernetes.io/instance: {{ .Release.Name }}

      initContainers:
        - name: wait-for-nacos
          image: "{{ .Values.global.busyboxImage.repository }}:{{ .Values.global.busyboxImage.tag }}"
          imagePullPolicy: IfNotPresent
          command:
            - sh
            - -c
            - |
              until nc -z ruoyi-nacos 8848; do
                echo "等待 Nacos"
                sleep 3
              done

        - name: wait-for-redis
          image: "{{ .Values.global.busyboxImage.repository }}:{{ .Values.global.busyboxImage.tag }}"
          imagePullPolicy: IfNotPresent
          command:
            - sh
            - -c
            - |
              until nc -z ruoyi-redis 6379; do
                echo "等待 Redis"
                sleep 3
              done

        - name: wait-for-mysql
          image: "{{ .Values.global.busyboxImage.repository }}:{{ .Values.global.busyboxImage.tag }}"
          imagePullPolicy: IfNotPresent
          command:
            - sh
            - -c
            - |
              until nc -z ruoyi-mysql 3306; do
                echo "等待 Mysql"
                sleep 3
              done

      containers:
        - name: system
          image: "{{ .Values.system.image.repository }}:{{ .Values.system.image.tag }}"
          imagePullPolicy: {{ .Values.global.imagePullPolicy }}

          env:
            - name: JAVA_OPTS
              value: {{ .Values.system.javaOpts | quote }}

            - name: SPRING_PROFILES_ACTIVE
              value: {{ .Values.global.profile | quote }}

            - name: MANAGEMENT_ENDPOINTS_WEB_EXPOSURE_INCLUDE
              value: "health,info,prometheus"

            - name: MANAGEMENT_ENDPOINT_PROMETHEUS_ENABLED
              value: "true"

            - name: MANAGEMENT_PROMETHEUS_METRICS_EXPORT_ENABLED
              value: "true"

            - name: MANAGEMENT_METRICS_TAGS_APPLICATION
              value: {{ .Values.system.name | quote }}

            - name: SPRING_CLOUD_NACOS_SERVER_ADDR
              value: {{ .Values.global.nacos.serverAddr | quote }}

            - name: SPRING_CLOUD_NACOS_CONFIG_SERVER_ADDR
              value: {{ .Values.global.nacos.serverAddr | quote }}

            - name: SPRING_CLOUD_NACOS_DISCOVERY_SERVER_ADDR
              value: {{ .Values.global.nacos.serverAddr | quote }}

            - name: SPRING_CLOUD_NACOS_USERNAME
              valueFrom:
                secretKeyRef:
                  name: ruoyi-app-nacos
                  key: username

            - name: SPRING_CLOUD_NACOS_PASSWORD
              valueFrom:
                secretKeyRef:
                  name: ruoyi-app-nacos
                  key: password

            - name: SPRING_CLOUD_NACOS_CONFIG_USERNAME
              valueFrom:
                secretKeyRef:
                  name: ruoyi-app-nacos
                  key: username

            - name: SPRING_CLOUD_NACOS_CONFIG_PASSWORD
              valueFrom:
                secretKeyRef:
                  name: ruoyi-app-nacos
                  key: password

            - name: SPRING_CLOUD_NACOS_DISCOVERY_USERNAME
              valueFrom:
                secretKeyRef:
                  name: ruoyi-app-nacos
                  key: username

            - name: SPRING_CLOUD_NACOS_DISCOVERY_PASSWORD
              valueFrom:
                secretKeyRef:
                  name: ruoyi-app-nacos
                  key: password

          ports:
            - name: http
              containerPort: {{ .Values.system.port }}

          startupProbe:
            tcpSocket:
              port: http
            periodSeconds: 5
            failureThreshold: 60

          readinessProbe:
            tcpSocket:
              port: http
            periodSeconds: 10
            failureThreshold: 6

          livenessProbe:
            tcpSocket:
              port: http
            periodSeconds: 20
            failureThreshold: 6

          resources:
{{ toYaml .Values.system.resources | indent 12 }}
EOF

十五、给三个 Service 添加监控标签

bash 复制代码 
cd "$HOME/ruoyi-k8s-lab/charts/ruoyi-app"

cat <<'EOF' > templates/services.yaml
apiVersion: v1
kind: Service
metadata:
  name: {{ .Values.gateway.name }}
  labels:
    app.kubernetes.io/name: {{ .Values.gateway.name }}
    app.kubernetes.io/instance: {{ .Release.Name }}
    monitoring: ruoyi-jvm
spec:
  type: ClusterIP
  selector:
    app.kubernetes.io/name: {{ .Values.gateway.name }}
    app.kubernetes.io/instance: {{ .Release.Name }}
  ports:
    - name: http
      port: {{ .Values.gateway.port }}
      targetPort: http
---
apiVersion: v1
kind: Service
metadata:
  name: {{ .Values.auth.name }}
  labels:
    app.kubernetes.io/name: {{ .Values.auth.name }}
    app.kubernetes.io/instance: {{ .Release.Name }}
    monitoring: ruoyi-jvm
spec:
  type: ClusterIP
  selector:
    app.kubernetes.io/name: {{ .Values.auth.name }}
    app.kubernetes.io/instance: {{ .Release.Name }}
  ports:
    - name: http
      port: {{ .Values.auth.port }}
      targetPort: http
---
apiVersion: v1
kind: Service
metadata:
  name: {{ .Values.system.name }}
  labels:
    app.kubernetes.io/name: {{ .Values.system.name }}
    app.kubernetes.io/instance: {{ .Release.Name }}
    monitoring: ruoyi-jvm
spec:
  type: ClusterIP
  selector:
    app.kubernetes.io/name: {{ .Values.system.name }}
    app.kubernetes.io/instance: {{ .Release.Name }}
  ports:
    - name: http
      port: {{ .Values.system.port }}
      targetPort: http
EOF

sed -i 's/^version:.*/version: 0.3.0/' Chart.yaml

十六、检查 Helm Chart

bash 复制代码 
cd "$HOME/ruoyi-k8s-lab"

helm lint \
  ./charts/ruoyi-app \
  -f ./charts/ruoyi-app/values-lab.yaml

必须显示：

text 复制代码 
1 chart(s) linted, 0 chart(s) failed

渲染：

bash 复制代码 
helm template ruoyi-app \
  ./charts/ruoyi-app \
  --namespace ruoyi \
  -f ./charts/ruoyi-app/values-lab.yaml \
  > /tmp/ruoyi-app-monitoring.yaml

检查：

bash 复制代码 
echo '===== 镜像 ====='
grep -n '3.6.8-monitoring-v1' /tmp/ruoyi-app-monitoring.yaml

echo
echo '===== 副本 ====='
grep -n 'replicas: 2' /tmp/ruoyi-app-monitoring.yaml

echo
echo '===== 监控环境变量 ====='
grep -nE \
  'MANAGEMENT_ENDPOINTS_WEB_EXPOSURE_INCLUDE|MANAGEMENT_ENDPOINT_PROMETHEUS_ENABLED|MANAGEMENT_PROMETHEUS_METRICS_EXPORT_ENABLED' \
  /tmp/ruoyi-app-monitoring.yaml

echo
echo '===== Service 标签 ====='
grep -n 'monitoring: ruoyi-jvm' /tmp/ruoyi-app-monitoring.yaml

echo
echo '===== 滚动策略与反亲和 ====='
grep -nE 'maxUnavailable|maxSurge|podAntiAffinity' \
  /tmp/ruoyi-app-monitoring.yaml

预期：

text 复制代码 
monitoring-v1 镜像出现 3 次
replicas: 2 出现 3 次
monitoring: ruoyi-jvm 出现 3 次
三个 Deployment 都有 Prometheus 环境变量、滚动策略和反亲和

十七、升级 ruoyi-app

bash 复制代码 
cd "$HOME/ruoyi-k8s-lab"

helm upgrade --install ruoyi-app \
  ./charts/ruoyi-app \
  --namespace ruoyi \
  -f ./charts/ruoyi-app/values-lab.yaml \
  --atomic \
  --wait \
  --timeout 20m

检查：

bash 复制代码 
helm list -n ruoyi
helm history ruoyi-app -n ruoyi

kubectl get deployment \
  ruoyi-gateway ruoyi-auth ruoyi-system \
  -n ruoyi

kubectl get pods \
  -n ruoyi \
  -l app.kubernetes.io/instance=ruoyi-app \
  -o wide

三个 Deployment 都应为 2/2。

十八、验证 Prometheus 指标端点

bash 复制代码 
kubectl run ruoyi-metrics-check \
  -n ruoyi \
  --rm -i \
  --restart=Never \
  --image=m.daocloud.io/docker.io/library/busybox:1.36.1 \
  -- sh -c '
    set -e
    for target in \
      ruoyi-gateway:8080 \
      ruoyi-auth:9200 \
      ruoyi-system:9201
    do
      echo
      echo "===== ${target} ====="
      wget -qO- "http://${target}/actuator/prometheus" | head -n 12
    done
  '

每个服务都应输出 JVM 指标。

检查 application 标签：

bash 复制代码 
kubectl run ruoyi-metrics-tag-check \
  -n ruoyi \
  --rm -i \
  --restart=Never \
  --image=m.daocloud.io/docker.io/library/busybox:1.36.1 \
  -- sh -c '
    wget -qO- \
      http://ruoyi-system:9201/actuator/prometheus \
      | grep "application=\"ruoyi-system\"" \
      | head
  '

十九、创建 ServiceMonitor

bash 复制代码 
mkdir -p "$HOME/ruoyi-k8s-lab/monitoring"

cat <<'EOF' > \
  "$HOME/ruoyi-k8s-lab/monitoring/ruoyi-servicemonitor.yaml"
apiVersion: monitoring.coreos.com/v1
kind: ServiceMonitor
metadata:
  name: ruoyi-jvm
  namespace: monitoring
  labels:
    app.kubernetes.io/part-of: ruoyi
spec:
  namespaceSelector:
    matchNames:
      - ruoyi
  selector:
    matchLabels:
      monitoring: ruoyi-jvm
  endpoints:
    - port: http
      path: /actuator/prometheus
      interval: 15s
      scrapeTimeout: 10s
      scheme: http
EOF

kubectl apply -f \
  "$HOME/ruoyi-k8s-lab/monitoring/ruoyi-servicemonitor.yaml"

自动匹配当前 Prometheus 的 ServiceMonitor 选择标签：

bash 复制代码 
PROMETHEUS_NAME="$(
  kubectl get prometheus \
    -n monitoring \
    -o jsonpath='{.items[0].metadata.name}'
)"

if [ -z "$PROMETHEUS_NAME" ]; then
  echo "没有找到 Prometheus 资源"
  exit 1
fi

echo "Prometheus：$PROMETHEUS_NAME"

SELECTOR_LABELS="$(
  kubectl get prometheus \
    "$PROMETHEUS_NAME" \
    -n monitoring \
    -o go-template='{{range $key, $value := .spec.serviceMonitorSelector.matchLabels}}{{printf "%s=%s\n" $key $value}}{{end}}'
)"

if [ -n "$SELECTOR_LABELS" ]; then
  printf '%s\n' "$SELECTOR_LABELS"

  while IFS='=' read -r key value; do
    [ -n "$key" ] || continue

    kubectl label servicemonitor \
      ruoyi-jvm \
      -n monitoring \
      "${key}=${value}" \
      --overwrite
  done <<< "$SELECTOR_LABELS"
else
  echo "Prometheus 未配置 matchLabels，无需添加额外标签"
fi

检查：

bash 复制代码 
kubectl get servicemonitor \
  ruoyi-jvm \
  -n monitoring \
  --show-labels

kubectl get svc \
  -n ruoyi \
  -l monitoring=ruoyi-jvm \
  --show-labels

二十、检查 Prometheus Target

bash 复制代码 
sleep 30

浏览器打开：

text 复制代码 
http://192.168.89.104:30031/targets

搜索 ruoyi，应看到：

text 复制代码 
Gateway 2 个 UP
Auth 2 个 UP
System 2 个 UP

API 查询：

bash 复制代码 
curl -sG \
  --data-urlencode \
  'query=count by (service) (up{namespace="ruoyi",service=~"ruoyi-(gateway|auth|system)"} == 1)' \
  http://192.168.89.104:30031/api/v1/query \
  | python3 -m json.tool

二十一、Grafana 查询

打开：

text 复制代码 
http://192.168.89.104:30030

进入 Explore → Prometheus。

Target 存活

promql 复制代码 
up{namespace="ruoyi",service=~"ruoyi-(gateway|auth|system)"}

JVM 堆内存

promql 复制代码 
sum by (application) (
  jvm_memory_used_bytes{namespace="ruoyi",area="heap"}
)

JVM 堆使用率

promql 复制代码 
100 *
sum by (application) (
  jvm_memory_used_bytes{namespace="ruoyi",area="heap"}
)
/
sum by (application) (
  jvm_memory_max_bytes{namespace="ruoyi",area="heap"}
)

JVM 活跃线程

promql 复制代码 
sum by (application) (
  jvm_threads_live_threads{namespace="ruoyi"}
)

进程 CPU

promql 复制代码 
avg by (application) (
  process_cpu_usage{namespace="ruoyi"}
) * 100

GC 暂停速率

promql 复制代码 
sum by (application) (
  rate(jvm_gc_pause_seconds_count{namespace="ruoyi"}[5m])
)

HTTP 请求速率

promql 复制代码 
sum by (application) (
  rate(http_server_requests_seconds_count{namespace="ruoyi"}[5m])
)

二十二、练习滚动升级到 monitoring-v2

22.1 创建 v2 标签并导出

bash 复制代码 
export OLD_TAG="3.6.8-monitoring-v1"
export NEW_TAG="3.6.8-monitoring-v2"
export ARTIFACT_DIR="$HOME/ruoyi-k8s-lab/artifacts"

for service in gateway auth system; do
  sudo buildah tag \
    "localhost/ruoyi-${service}:${OLD_TAG}" \
    "localhost/ruoyi-${service}:${NEW_TAG}"

  rm -f "${ARTIFACT_DIR}/ruoyi-${service}-${NEW_TAG}.tar"

  sudo buildah push \
    "localhost/ruoyi-${service}:${NEW_TAG}" \
    "docker-archive:${ARTIFACT_DIR}/ruoyi-${service}-${NEW_TAG}.tar:localhost/ruoyi-${service}:${NEW_TAG}"
done

sudo chown leecurry:leecurry \
  "${ARTIFACT_DIR}/ruoyi-gateway-${NEW_TAG}.tar" \
  "${ARTIFACT_DIR}/ruoyi-auth-${NEW_TAG}.tar" \
  "${ARTIFACT_DIR}/ruoyi-system-${NEW_TAG}.tar"

22.2 逐台发送

bash 复制代码 
scp "${ARTIFACT_DIR}"/*"${NEW_TAG}".tar \
  leecurry@192.168.89.104:/tmp/
bash 复制代码 
scp "${ARTIFACT_DIR}"/*"${NEW_TAG}".tar \
  leecurry@192.168.89.105:/tmp/
bash 复制代码 
scp "${ARTIFACT_DIR}"/*"${NEW_TAG}".tar \
  leecurry@192.168.89.106:/tmp/

22.3 三台 Worker 分别导入

每台 Worker 都执行：

bash 复制代码 
export NEW_TAG="3.6.8-monitoring-v2"

for image_file in \
  "/tmp/ruoyi-gateway-${NEW_TAG}.tar" \
  "/tmp/ruoyi-auth-${NEW_TAG}.tar" \
  "/tmp/ruoyi-system-${NEW_TAG}.tar"
do
  [ -s "$image_file" ] || {
    echo "文件不存在或为空：$image_file"
    exit 1
  }

  sudo ctr -n k8s.io images import "$image_file" || exit 1
done

sudo crictl images \
  | grep -E 'ruoyi-(gateway|auth|system)' \
  | grep "$NEW_TAG"

22.4 修改 Helm 标签并升级

bash 复制代码 
cd "$HOME/ruoyi-k8s-lab/charts/ruoyi-app"

sed -i \
  's/3\.6\.8-monitoring-v1/3.6.8-monitoring-v2/g' \
  values-lab.yaml

grep -n '3.6.8-monitoring-v2' values-lab.yaml

另开窗口观察：

bash 复制代码 
watch -n 1 \
  kubectl get pods \
  -n ruoyi \
  -l app.kubernetes.io/instance=ruoyi-app \
  -o wide

原窗口升级：

bash 复制代码 
cd "$HOME/ruoyi-k8s-lab"

helm upgrade ruoyi-app \
  ./charts/ruoyi-app \
  --namespace ruoyi \
  -f ./charts/ruoyi-app/values-lab.yaml \
  --atomic \
  --wait \
  --timeout 20m

检查：

bash 复制代码 
kubectl rollout status deployment/ruoyi-gateway \
  -n ruoyi --timeout=10m
kubectl rollout status deployment/ruoyi-auth \
  -n ruoyi --timeout=10m
kubectl rollout status deployment/ruoyi-system \
  -n ruoyi --timeout=10m

kubectl get deployment \
  -n ruoyi \
  -o custom-columns='NAME:.metadata.name,IMAGES:.spec.template.spec.containers[*].image' \
  | grep -E 'ruoyi-(gateway|auth|system)'

二十三、练习删除 Pod 后自动恢复

bash 复制代码 
GATEWAY_POD="$(
  kubectl get pods \
    -n ruoyi \
    -l app.kubernetes.io/name=ruoyi-gateway \
    -o jsonpath='{.items[0].metadata.name}'
)"

echo "$GATEWAY_POD"
kubectl delete pod "$GATEWAY_POD" -n ruoyi

kubectl get pods \
  -n ruoyi \
  -l app.kubernetes.io/name=ruoyi-gateway \
  -o wide \
  -w

重新恢复两个 1/1 Running 后按 Ctrl + C。

二十四、练习 Java 进程崩溃后的容器重启

bash 复制代码 
AUTH_POD="$(
  kubectl get pods \
    -n ruoyi \
    -l app.kubernetes.io/name=ruoyi-auth \
    -o jsonpath='{.items[0].metadata.name}'
)"

echo "$AUTH_POD"

kubectl get pod "$AUTH_POD" -n ruoyi \
  -o custom-columns='NAME:.metadata.name,RESTARTS:.status.containerStatuses[0].restartCount'

kubectl exec -n ruoyi "$AUTH_POD" \
  -- sh -c 'kill -9 1' || true

sleep 20

kubectl get pod "$AUTH_POD" -n ruoyi \
  -o custom-columns='NAME:.metadata.name,STATUS:.status.phase,READY:.status.containerStatuses[0].ready,RESTARTS:.status.containerStatuses[0].restartCount'

RESTARTS 应增加，Pod 最终重新 Ready。

二十五、安全练习节点不可调度后的迁移

MySQL、Redis、Nacos 使用 local-path，不要对 Worker 直接执行 drain，也不要关闭承载本地 PVC 的节点。

bash 复制代码 
GATEWAY_POD="$(
  kubectl get pods \
    -n ruoyi \
    -l app.kubernetes.io/name=ruoyi-gateway \
    -o jsonpath='{.items[0].metadata.name}'
)"

GATEWAY_NODE="$(
  kubectl get pod "$GATEWAY_POD" \
    -n ruoyi \
    -o jsonpath='{.spec.nodeName}'
)"

echo "Pod：${GATEWAY_POD}"
echo "节点：${GATEWAY_NODE}"

kubectl cordon "$GATEWAY_NODE"
kubectl delete pod "$GATEWAY_POD" -n ruoyi

kubectl get pods \
  -n ruoyi \
  -l app.kubernetes.io/name=ruoyi-gateway \
  -o wide \
  -w

新 Pod 在其他 Worker 启动后按 Ctrl + C，然后恢复节点：

bash 复制代码 
kubectl uncordon "$GATEWAY_NODE"
kubectl get nodes

二十六、清理 Worker 临时 tar

确认 monitoring-v2 正常运行后，在三个 Worker 分别执行：

bash 复制代码 
sudo rm -f \
  /tmp/ruoyi-gateway-3.6.8-monitoring-v1.tar \
  /tmp/ruoyi-auth-3.6.8-monitoring-v1.tar \
  /tmp/ruoyi-system-3.6.8-monitoring-v1.tar \
  /tmp/ruoyi-gateway-3.6.8-monitoring-v2.tar \
  /tmp/ruoyi-auth-3.6.8-monitoring-v2.tar \
  /tmp/ruoyi-system-3.6.8-monitoring-v2.tar

二十七、最终检查

bash 复制代码 
echo '===== Helm ====='
helm list -n ruoyi

echo
echo '===== Deployment ====='
kubectl get deployment \
  ruoyi-gateway ruoyi-auth ruoyi-system \
  -n ruoyi

echo
echo '===== Pod ====='
kubectl get pods -n ruoyi -o wide

echo
echo '===== ServiceMonitor ====='
kubectl get servicemonitor \
  ruoyi-jvm \
  -n monitoring \
  --show-labels

echo
echo '===== 监控 Service ====='
kubectl get svc \
  -n ruoyi \
  -l monitoring=ruoyi-jvm

echo
echo '===== 节点 ====='
kubectl get nodes

最终标准：

text 复制代码 
Gateway：2/2
Auth：2/2
System：2/2
Gateway/Auth/System 镜像：3.6.8-monitoring-v2
Prometheus 中每个服务有 2 个 UP Target
Grafana 可以查询 JVM 堆、线程、CPU、GC、HTTP 指标
删除 Pod 后自动补回
Java 进程退出后容器自动重启
cordon 后无状态 Pod 可以迁移到其他 Worker
所有节点最终恢复 Ready，没有 SchedulingDisabled

二十八、回滚命令

查看 Helm 历史：

bash 复制代码 
helm history ruoyi-app -n ruoyi

回滚：

bash 复制代码 
helm rollback ruoyi-app <上一版本Revision> \
  -n ruoyi \
  --wait \
  --timeout 20m

查看 Deployment 历史：

bash 复制代码 
kubectl rollout history deployment/ruoyi-gateway -n ruoyi
kubectl rollout history deployment/ruoyi-auth -n ruoyi
kubectl rollout history deployment/ruoyi-system -n ruoyi
相关推荐
CoderYanger1 小时前
Java EE：9.JVM（课件内容-上篇）
java·jvm·程序人生·面试·职场和发展·java-ee·程序员创富
残月心殇 请珍惜枸1 小时前
.NET陷阱之五：奇怪的OutOfMemoryException——大对象堆引起的问题与对策
java·jvm·.net
mqiqe2 小时前
构建企业级 AI 原生架构：LLM Gateway、RAG、Agent 与 MCP 的协同关系解析
人工智能·架构·gateway
CoderYanger2 小时前
Java EE：9.JVM（课件内容-下篇）
java·jvm·程序人生·面试·职场和发展·java-ee·学习方法
mqiqe13 小时前
LLM Gateway：企业级大模型应用的“统一控制平面”
平面·flask·gateway
八角.。1 天前
方法参数与Debug按键
java·开发语言·jvm
晨陌y1 天前
redis_exporter + Prometheus监控Redis: 指标采集、告警与远程抓取
redis·bootstrap·prometheus
敲代码的嘎仔1 天前
从零搭建微服务：Spring Cloud Alibaba 全家桶踩坑实录（Nacos + OpenFeign + Gateway + Sentinel）
java·spring boot·spring·spring cloud·微服务·gateway·sentinel
Ming_studying2 天前
Python + SQLite FTS5 构建本地文档全文搜索器：增量索引、中文检索与高亮
jvm·数据库·python·sqlite
热门推荐
01如何新建文件夹？ 电脑新建文件夹的4种方法02GitHub 镜像站点03国内可直接用、免费额度/永久免费的大模型API清单（含 SiliconFlow、火山、阿里、智谱、百度、Kimi、DeepSeek、DMXAPI 等）04AI 编程 IDE 全景解析 2026：Agent 全面接管开发链路05Agnes AI 免费 API 接入指南：文本、生图、生视频，一套接口全免费06微信历史版本含下载地址（ Windows PC | 安卓 | MAC ）及设置微信不更新072026 国产 AI 大模型横评：DeepSeek、通义千问、Kimi、文心一言、星火、豆包谁更能打？082026年7月AI圈大地震：GPT-5.6被政府限制、Claude入驻Slack、Anthropic自研芯片09历年考研数学一、数学二、数学三真题试卷及答案PDF102026 年 AI 大模型 & AI 编程工具实战全总结