Rabbit mq - 等待 Mnesia 表时出错

Question

我在 kubernetes 集群上使用 helm chart 安装了 rabbitmq。 rabbitmq pod 不断重启。在检查 pod 日志时，我收到以下错误

2020-02-26 04:42:31.582 [warning] <0.314.0> Error while waiting for Mnesia tables: {timeout_waiting_for_tables,[rabbit_durable_queue]}
2020-02-26 04:42:31.582 [info] <0.314.0> Waiting for Mnesia tables for 30000 ms, 6 retries left

当我尝试执行 kubectl describe pod 时出现此错误

Conditions:
  Type              Status
  Initialized       True
  Ready             False
  ContainersReady   False
  PodScheduled      True
Volumes:
  data:
    Type:       PersistentVolumeClaim (a reference to a PersistentVolumeClaim in the same namespace)
    ClaimName:  data-rabbitmq-0
    ReadOnly:   false
  config-volume:
    Type:      ConfigMap (a volume populated by a ConfigMap)
    Name:      rabbitmq-config
    Optional:  false
  healthchecks:
    Type:      ConfigMap (a volume populated by a ConfigMap)
    Name:      rabbitmq-healthchecks
    Optional:  false
  rabbitmq-token-w74kb:
    Type:        Secret (a volume populated by a Secret)
    SecretName:  rabbitmq-token-w74kb
    Optional:    false
QoS Class:       Burstable
Node-Selectors:  beta.kubernetes.io/arch=amd64
Tolerations:     node.kubernetes.io/not-ready:NoExecute for 300s
                 node.kubernetes.io/unreachable:NoExecute for 300s
Events:
  Type     Reason     Age                      From                                               Message
  ----     ------     ----                     ----                                               -------
  Warning  Unhealthy  3m27s (x878 over 7h21m)  kubelet, gke-analytics-default-pool-918f5943-w0t0  Readiness probe failed: Timeout: 70 seconds ...
Checking health of node rabbit@rabbitmq-0.rabbitmq-headless.default.svc.cluster.local ...
Status of node rabbit@rabbitmq-0.rabbitmq-headless.default.svc.cluster.local ...
Error:
{:aborted, {:no_exists, [:rabbit_vhost, [{{:vhost, :"$1", :_, :_}, [], [:"$1"]}]]}}
Error:
{:aborted, {:no_exists, [:rabbit_vhost, [{{:vhost, :"$1", :_, :_}, [], [:"$1"]}]]}}

我已在 Kubernetes 集群上的 Google Cloud 上配置了上述内容。我不确定它是在什么特定情况下开始失败的。我不得不重新启动 pod，从那以后它一直失败。

这里有什么问题？

Answer 1

就我而言，解决方案很简单

Step1：缩小有状态集，它不会删除 PVC。

kubectl scale statefulsets rabbitmq-1-rabbitmq --namespace teps-rabbitmq --replicas=1

第二步：访问 RabbitMQ Pod。

kubectl exec -it rabbitmq-1-rabbitmq-0 -n Rabbit

Step3：重置集群

rabbitmqctl stop_app
rabbitmqctl force_boot

Step4：重新调整有状态集

  kubectl scale statefulsets rabbitmq-1-rabbitmq --namespace teps-rabbitmq --replicas=4

Answer 2

TLDR

helm upgrade rabbitmq --set clustering.forceBoot=true

问题

出现问题的原因如下：

• 由于某种原因，所有 RMQ pod 同时终止（可能是因为您将 StatefulSet 副本显式设置为 0，或其他原因）
• 其中一个是最后一个停止的（可能只是在其他几个之后一点点）。它将这个条件（“我现在是独立的”）存储在它的文件系统中，在 k8s 中是 PersistentVolume(Claim)。假设这个 pod 是 rabbitmq-1。
• 当您重新启动 StatefulSet 时，pod rabbitmq-0 始终是第一个启动的（请参阅 here）。
• 在启动过程中，pod rabbitmq-0 首先检查它是否应该独立运行。但就它在自己的文件系统上看到的而言，它是集群的一部分。所以它检查它的对等点，但没有找到。这个results in a startup failure by default。
• rabbitmq-0 因此永远不会准备好。
• rabbitmq-1 永远不会启动，因为这就是 StatefulSet 的部署方式——一个接一个。如果它要启动，它将成功启动，因为它看到它也可以独立运行。

所以说到底，RabbitMQ 和 StatefulSets 的工作方式有点不匹配。 RMQ 说：“如果一切都失败了，只需启动一切，同时，一个就能启动，一旦这个启动，其他人就可以重新加入集群。” k8s StatefulSets 说：“一次启动所有东西是不可能的，我们将从 0 开始”。

解决方案

为了解决这个问题，rabbitmqctl 有一个force_boot 命令，它基本上告诉实例在找不到任何对等点时独立启动。如何在 Kubernetes 中使用它取决于您使用的 Helm 图表和容器。在使用Bitnami Docker image 的Bitnami Chart 中，有一个值clustering.forceBoot = true，它转换为容器中的环境变量RABBITMQ_FORCE_BOOT = yes，然后容器将为您发出上述命令。

但是看看这个问题，你也可以看到为什么删除 PVC 会起作用 (other answer)。 pod 将完全“忘记”它们上次是 RMQ 集群的一部分，并愉快地开始。不过，我更喜欢上述解决方案，因为不会丢失任何数据。

Answer 3

测试这个部署：

kind: Service
apiVersion: v1
metadata:
  namespace: rabbitmq-namespace
  name: rabbitmq
  labels:
    app: rabbitmq
    type: LoadBalancer  
spec:
  type: NodePort
  ports:
   - name: http
     protocol: TCP
     port: 15672
     targetPort: 15672
     nodePort: 31672
   - name: amqp
     protocol: TCP
     port: 5672
     targetPort: 5672
     nodePort: 30672
   - name: stomp
     protocol: TCP
     port: 61613
     targetPort: 61613
  selector:
    app: rabbitmq
---
kind: Service 
apiVersion: v1
metadata:
  namespace: rabbitmq-namespace
  name: rabbitmq-lb
  labels:
    app: rabbitmq
spec:
  # Headless service to give the StatefulSet a DNS which is known in the cluster (hostname-#.app.namespace.svc.cluster.local, )
  # in our case - rabbitmq-#.rabbitmq.rabbitmq-namespace.svc.cluster.local  
  clusterIP: None
  ports:
   - name: http
     protocol: TCP
     port: 15672
     targetPort: 15672
   - name: amqp
     protocol: TCP
     port: 5672
     targetPort: 5672
   - name: stomp
     port: 61613
  selector:
    app: rabbitmq
---
apiVersion: v1
kind: ConfigMap
metadata:
  name: rabbitmq-config
  namespace: rabbitmq-namespace
data:
  enabled_plugins: |
      [rabbitmq_management,rabbitmq_peer_discovery_k8s,rabbitmq_stomp].

  rabbitmq.conf: |
      ## Cluster formation. See http://www.rabbitmq.com/cluster-formation.html to learn more.
      cluster_formation.peer_discovery_backend  = rabbit_peer_discovery_k8s
      cluster_formation.k8s.host = kubernetes.default.svc.cluster.local
      ## Should RabbitMQ node name be computed from the pod's hostname or IP address?
      ## IP addresses are not stable, so using [stable] hostnames is recommended when possible.
      ## Set to "hostname" to use pod hostnames.
      ## When this value is changed, so should the variable used to set the RABBITMQ_NODENAME
      ## environment variable.
      cluster_formation.k8s.address_type = hostname   
      ## Important - this is the suffix of the hostname, as each node gets "rabbitmq-#", we need to tell what's the suffix
      ## it will give each new node that enters the way to contact the other peer node and join the cluster (if using hostname)
      cluster_formation.k8s.hostname_suffix = .rabbitmq.rabbitmq-namespace.svc.cluster.local
      ## How often should node cleanup checks run?
      cluster_formation.node_cleanup.interval = 30
      ## Set to false if automatic removal of unknown/absent nodes
      ## is desired. This can be dangerous, see
      ##  * http://www.rabbitmq.com/cluster-formation.html#node-health-checks-and-cleanup
      ##  * https://groups.google.com/forum/#!msg/rabbitmq-users/wuOfzEywHXo/k8z_HWIkBgAJ
      cluster_formation.node_cleanup.only_log_warning = true
      cluster_partition_handling = autoheal
      ## See http://www.rabbitmq.com/ha.html#master-migration-data-locality
      queue_master_locator=min-masters
      ## See http://www.rabbitmq.com/access-control.html#loopback-users
      loopback_users.guest = false
---
apiVersion: apps/v1
kind: StatefulSet
metadata:
  name: rabbitmq
  namespace: rabbitmq-namespace
spec:
  serviceName: rabbitmq
  replicas: 3
  selector:
    matchLabels:
      name: rabbitmq
  template:
    metadata:
      labels:
        app: rabbitmq
        name: rabbitmq
        state: rabbitmq
      annotations:
        pod.alpha.kubernetes.io/initialized: "true"
    spec:
      serviceAccountName: rabbitmq
      terminationGracePeriodSeconds: 10
      containers:        
      - name: rabbitmq-k8s
        image: rabbitmq:3.8.3
        volumeMounts:
          - name: config-volume
            mountPath: /etc/rabbitmq
          - name: data
            mountPath: /var/lib/rabbitmq/mnesia
        ports:
          - name: http
            protocol: TCP
            containerPort: 15672
          - name: amqp
            protocol: TCP
            containerPort: 5672
        livenessProbe:
          exec:
            command: ["rabbitmqctl", "status"]
          initialDelaySeconds: 60
          periodSeconds: 60
          timeoutSeconds: 10
        resources:
            requests:
              memory: "0"
              cpu: "0"
            limits:
              memory: "2048Mi"
              cpu: "1000m"
        readinessProbe:
          exec:
            command: ["rabbitmqctl", "status"]
          initialDelaySeconds: 20
          periodSeconds: 60
          timeoutSeconds: 10
        imagePullPolicy: Always
        env:
          - name: MY_POD_IP
            valueFrom:
              fieldRef:
                fieldPath: status.podIP
          - name: NAMESPACE
            valueFrom:
              fieldRef:
                fieldPath: metadata.namespace
          - name: HOSTNAME
            valueFrom:
              fieldRef:
                fieldPath: metadata.name
          - name: RABBITMQ_USE_LONGNAME
            value: "true"
          # See a note on cluster_formation.k8s.address_type in the config file section
          - name: RABBITMQ_NODENAME
            value: "rabbit@$(HOSTNAME).rabbitmq.$(NAMESPACE).svc.cluster.local"
          - name: K8S_SERVICE_NAME
            value: "rabbitmq"
          - name: RABBITMQ_ERLANG_COOKIE
            value: "mycookie"      
      volumes:
        - name: config-volume
          configMap:
            name: rabbitmq-config
            items:
            - key: rabbitmq.conf
              path: rabbitmq.conf
            - key: enabled_plugins
              path: enabled_plugins
  volumeClaimTemplates:
  - metadata:
      name: data
    spec:
      accessModes:
        - "ReadWriteOnce"
      storageClassName: "default"
      resources:
        requests:
          storage: 3Gi

---
apiVersion: v1
kind: ServiceAccount
metadata:
  name: rabbitmq 
  namespace: rabbitmq-namespace 
---
kind: Role
apiVersion: rbac.authorization.k8s.io/v1beta1
metadata:
  name: endpoint-reader
  namespace: rabbitmq-namespace 
rules:
- apiGroups: [""]
  resources: ["endpoints"]
  verbs: ["get"]
---
kind: RoleBinding
apiVersion: rbac.authorization.k8s.io/v1beta1
metadata:
  name: endpoint-reader
  namespace: rabbitmq-namespace
subjects:
- kind: ServiceAccount
  name: rabbitmq
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: Role
  name: endpoint-reader

Answer 4

我也遇到了类似的错误，如下所示。

2020-06-05 03:45:37.153 [info] 等待 Mnesia 表 30000 毫秒，还剩 9 次重试 2020-06-05 03:46:07.154 [警告] 等待 Mnesia 表时出错： {timeout_waiting_for_tables,[rabbit_user,rabbit_user_permission,rabbit_topic_permission,rabbit_vhost,rabbit_durable_route,rabbit_durable_exchange,rabbit_runtime_parameters,rabbit_durable_queue]} 2020-06-05 03:46:07.154 [info] 等待 Mnesia 表 30000 毫秒，还剩 8 次重试

在我的例子中，RabbitMQ 集群的从节点（服务器）宕机了。一旦我启动了从节点，主节点就启动了，没有错误。

Answer 5

刚刚删除了现有的持久卷声明并重新安装了rabbitmq，它开始工作了。

因此，每次在 kubernetes 集群上安装 rabbitmq 之后，如果我将 pod 缩小到 0 并且稍后扩大 pod 时，我都会遇到同样的错误。我还尝试在不卸载 rabbitmq helm 图表的情况下删除 Persistent Volume Claim，但仍然出现相同的错误。

所以看来每次我将集群缩小到0，都需要卸载rabbitmq helm chart，删除对应的Persistent Volume Claims，每次都安装rabbitmq helm chart才能正常工作。