部署介绍
在Kubernetes上搭建RabbitMQ有4种部署方法（IP模式、Pod与server的DNS、Statefulset 与Headless Service、hostname模式），以下是通过hostname模式来部署，经过多少测试使用yaml部署高可用RabbitMQ如下：
1.hostname模式部署yaml文件：
1 apiVersion: v1 2 kind: Namespace 3 metadata: 4 name: rabbitmq 5 --- 6 apiVersion: v1 7 kind: ServiceAccount 8 metadata: 9 name: rabbitmq 10 namespace: rabbitmq 11 --- 12 kind: Role 13 apiVersion: rbac.authorization.k8s.io/v1beta1 14 metadata: 15 name: endpoint-reader 16 namespace: rabbitmq 17 rules: 18 - apiGroups: [""] 19 resources: ["endpoints"] 20 verbs: ["get"] 21 --- 22 kind: RoleBinding 23 apiVersion: rbac.authorization.k8s.io/v1beta1 24 metadata: 25 name: endpoint-reader 26 namespace: rabbitmq 27 subjects: 28 - kind: ServiceAccount 29 name: rabbitmq 30 roleRef: 31 apiGroup: rbac.authorization.k8s.io 32 kind: Role 33 name: endpoint-reader 34 --- 35 apiVersion: v1 36 kind: PersistentVolume 37 metadata: 38 name: rabbitmq-data 39 labels: 40 release: rabbitmq-data 41 namespace: rabbitmq 42 spec: 43 capacity: 44 storage: 2Gi 45 accessModes: 46 - ReadWriteMany 47 persistentVolumeReclaimPolicy: Retain 48 nfs: 49 path: /opt/rabbitmq 50 server: 192.168.163.130 51 --- 52 apiVersion: v1 53 kind: PersistentVolumeClaim 54 metadata: 55 name: rabbitmq-data-claim 56 namespace: rabbitmq 57 spec: 58 accessModes: 59 - ReadWriteMany 60 resources: 61 requests: 62 storage: 2Gi 63 selector: 64 matchLabels: 65 release: rabbitmq-data 66 --- 67 kind: Service 68 apiVersion: v1 69 metadata: 70 name: rabbitmq-headless 71 namespace: rabbitmq 72 spec: 73 clusterIP: None 74 publishNotReadyAddresses: true 75 ports: 76 - name: amqp 77 port: 5672 78 - name: http 79 port: 15672 80 selector: 81 app: rabbitmq 82 --- 83 kind: Service 84 apiVersion: v1 85 metadata: 86 namespace: rabbitmq 87 name: rabbitmq-service 88 spec: 89 ports: 90 - name: http 91 protocol: TCP 92 port: 15672 93 targetPort: 15672 94 nodePort: 32001 95 - name: amqp 96 protocol: TCP 97 port: 5672 98 targetPort: 5672 99 nodeport 100 nodePort: 32002 101 selector: 102 app: rabbitmq 103 type: NodePort 104 --- 105 apiVersion: v1 106 kind: ConfigMap 107 metadata: 108 name: rabbitmq-config 109 namespace: rabbitmq 110 data: 111 enabled_plugins: | 112 [rabbitmq_management,rabbitmq_peer_discovery_k8s]. 113 rabbitmq.conf: | 114 cluster_formation.peer_discovery_backend = rabbit_peer_discovery_k8s 115 cluster_formation.k8s.host = kubernetes.default.svc.cluster.local 116 cluster_formation.k8s.address_type = hostname 117 cluster_formation.node_cleanup.interval = 10 118 cluster_formation.node_cleanup.only_log_warning = true 119 cluster_partition_handling = autoheal 120 queue_master_locator=min-masters 121 loopback_users.guest = false 122 123 cluster_formation.randomized_startup_delay_range.min = 0 124 cluster_formation.randomized_startup_delay_range.max = 2 125 cluster_formation.k8s.service_name = rabbitmq-headless 126 cluster_formation.k8s.hostname_suffix = .rabbitmq-headless.rabbitmq.svc.cluster.local 127 vm_memory_high_watermark.absolute = 1GB 128 disk_free_limit.absolute = 2GB 129 --- 130 apiVersion: apps/v1 131 kind: StatefulSet 132 metadata: 133 name: rabbitmq 134 namespace: rabbitmq 135 spec: 136 serviceName: rabbitmq-headless 137 selector: 138 matchLabels: 139 app: rabbitmq 140 replicas: 2 141 template: 142 metadata: 143 labels: 144 app: rabbitmq 145 annotations: 146 scheduler.alpha.kubernetes.io/affinity: > 147 { 148 "podAntiAffinity": { 149 "requiredDuringSchedulingIgnoredDuringExecution": [{ 150 "labelSelector": { 151 "matchExpressions": [{ 152 "key": "app", 153 "operator": "In", 154 "values": ["rabbitmq"] 155 }] 156 }, 157 "topologyKey": "kubernetes.io/hostname" 158 }] 159 } 160 } 161 spec: 162 serviceAccountName: rabbitmq 163 terminationGracePeriodSeconds: 10 164 containers: 165 - name: rabbitmq 166 image: 192.168.163.129:60080/rabbitmq:3.7.19-management 167 resources: 168 limits: 169 cpu: 1 170 memory: 1Gi 171 requests: 172 cpu: 1 173 memory: 1Gi 174 volumeMounts: 175 - name: config-volume 176 mountPath: /etc/rabbitmq 177 - name: rabbitmq-data 178 mountPath: /var/lib/rabbitmq/mnesia 179 ports: 180 - name: http 181 protocol: TCP 182 containerPort: 15672 183 - name: amqp 184 protocol: TCP 185 containerPort: 5672 186 livenessProbe: 187 exec: 188 command: ["rabbitmqctl", "status"] 189 initialDelaySeconds: 60 190 periodSeconds: 60 191 timeoutSeconds: 5 192 readinessProbe: 193 exec: 194 command: ["rabbitmqctl", "status"] 195 initialDelaySeconds: 20 196 periodSeconds: 60 197 timeoutSeconds: 5 198 imagePullPolicy: Always 199 env: 200 - name: HOSTNAME 201 valueFrom: 202 fieldRef: 203 fieldPath: metadata.name 204 - name: RABBITMQ_USE_LONGNAME 205 value: "true" 206 - name: RABBITMQ_NODENAME 207 value: "rabbit@$(HOSTNAME).rabbitmq-headless.rabbitmq.svc.cluster.local" 208 - name: RABBITMQ_ERLANG_COOKIE 209 value: "mycookie" 210 volumes: 211 - name: config-volume 212 configMap: 213 name: rabbitmq-config 214 items: 215 - key: rabbitmq.conf 216 path: rabbitmq.conf 217 - key: enabled_plugins 218 path: enabled_plugins 219 - name: rabbitmq-data 220 persistentVolumeClaim: 221 claimName: rabbitmq-data-claim
2.因官网给出的镜像无法实现镜像模式和自启动需要对镜像进行重新编译
dockerfile如下：
1 FROM paasprod01.ceair.com:31104/bpm/rabbitmq:3.7.19-management 2 3 copy rabbitmq.sh /root 4 RUN chmod 777 /root/rabbitmq.sh 5 CMD ["bash", "/root/rabbitmq.sh"]
启动脚本如下：
1 #!/bin/bash 2 3 rabbitmq-server & 4 sleep 30 5 rabbitmqctl set_policy ha-all "^" '{"ha-mode":"all"}' 6 tail -f /var/log/rabbitmq/log/crash.log
至此，终于在Kubernetes上部署完成