在 kubernetes 中远程访问 Kafka 运行

Remotely accessing Kafka running inside kubernetes

我在单节点 kubernetes 环境的 pod 中有一个单节点 Kafka 代理 运行。我将此图像用于 kafka:https://hub.docker.com/r/wurstmeister/kafka

kafka 版本 = 1.1.0

Kubernetes 集群 运行 在服务器上的虚拟机中。 VM 在活动接口 ens32 上具有以下 IP - 192.168.3.102

Kafka.yaml

apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  namespace: casb-deployment
  name: kafkaservice
spec:
  replicas: 1
  template:
    metadata:
      labels:
        app: kafkaservice
    spec:
      hostname: kafkaservice
      containers:
      - name: kafkaservice
        imagePullPolicy: IfNotPresent
        image: wurstmeister/kafka:1.1.0
        env:
         - name: KAFKA_BROKER_ID
           value: "1"
#         - name: KAFKA_ADVERTISED_HOST_NAME
#           value: "kafkaservice"
         - name: KAFKA_LISTENER_SECURITY_PROTOCOL_MAP
           value: "INTERNAL_PLAINTEXT:PLAINTEXT,EXTERNAL_PLAINTEXT:PLAINTEXT"
         - name: KAFKA_LISTENERS
           value: "INTERNAL_PLAINTEXT://0.0.0.0:9092,EXTERNAL_PLAINTEXT://0.0.0.0:9093"
         - name: KAFKA_ADVERTISED_LISTENERS
           value: "INTERNAL_PLAINTEXT://kafkaservice:9092,EXTERNAL_PLAINTEXT://192.168.3.102:9093"
         - name: KAFKA_INTER_BROKER_LISTENER_NAME
           value: "INTERNAL_PLAINTEXT"
         - name: KAFKA_CREATE_TOPICS
           value: "topic-1:100:1,topic-2:1:1"  
         - name: KAFKA_ZOOKEEPER_CONNECT
           value: "zookeeper:2181"
        ports: 
        - name: port9092
          containerPort: 9092
        - name: port9093
          containerPort: 9093
        volumeMounts:
        - mountPath: /kafka/kafka-logs-kafkaservice
          name: kafka-volume
      volumes:
      - name: kafka-volume
        hostPath:
          path: /home/volume/kafka-logs

---

apiVersion: v1
kind: Service
metadata:
  namespace: casb-deployment
  name: kafkaservice
  labels:
    app: kafkaservice
spec:
  selector:
    app: kafkaservice
  ports:
  - name: port9092
    port: 9092
    targetPort: 9092
    protocol: TCP

---

apiVersion: v1
kind: Service
metadata:
  namespace: casb-deployment
  name: kafkaservice-external
  labels:
    app: kafkaservice-external
spec:
  selector:
    app: kafkaservice
  ports:
  - name: port9093
    port: 9093
    protocol: TCP
    nodePort: 30035
  type: NodePort

我能够从我的本地计算机 ping 虚拟机,即 kubernetes 节点 ping 192.168.3.102,并且我正在使用 nodePort 公开服务。

我也可以远程登录 telnet 192.168.3.102 30035 并且它给出:

Trying 192.168.3.102...
Connected to 192.168.3.102.
Escape character is '^]'.

我在我的本地机器上尝试了 运行 kafka 控制台消费者和生产者:

消费者: bin/kafka-console-consumer.sh --bootstrap-server 192.168.3.102:30035 --topic topic-1 --from-beginning

输出:

[2019-09-25 12:30:40,716] WARN [Consumer clientId=consumer-1, groupId=console-consumer-20551] Connection to node 1 could not be established. Broker may not be available. (org.apache.kafka.clients.NetworkClient)

制作人:

bin/kafka-console-producer.sh --broker-list 192.168.3.102:30035 --topic topic-1

输出:

[2019-09-25 12:32:07,958] WARN [Producer clientId=console-producer] Connection to node 1 could not be established. Broker may not be available. (org.apache.kafka.clients.NetworkClient)

网络统计:

netstat -tulpn | grep 30035
tcp6       0      0 :::30035                :::*                    LISTEN      113545/kube-proxy

我尝试了 运行 基于 python 的消费者,即 kafka-python==1.4.2 它给了我以下日志:

[2019-09-25T12:15:39+0500] INFO kafka.client Bootstrapping cluster metadata from [('192.168.3.102', 30035, <AddressFamily.AF_INET: 2>)]
[2019-09-25T12:15:39+0500] INFO kafka.conn <BrokerConnection node_id=bootstrap host=192.168.3.102:30035 <connecting> [IPv4 ('192.168.3.102', 30035)]>: connecting to 192.168.3.102:30035 [('192.168.3.102', 30035) IPv4]
[2019-09-25T12:15:39+0500] INFO kafka.conn <BrokerConnection node_id=bootstrap host=192.168.3.102:30035 <connecting> [IPv4 ('192.168.3.102', 30035)]>: Connection complete.
[2019-09-25T12:15:39+0500] INFO kafka.client Bootstrap succeeded: found 1 brokers and 26 topics.
[2019-09-25T12:15:39+0500] INFO kafka.conn <BrokerConnection node_id=bootstrap host=192.168.3.102:30035 <connected> [IPv4 ('192.168.3.102', 30035)]>: Closing connection. 
[2019-09-25T12:15:39+0500] INFO kafka.conn <BrokerConnection node_id=1 host=192.168.3.102:9093 <connecting> [IPv4 ('192.168.3.102', 9093)]>: connecting to 192.168.3.102:9093 [('192.168.3.102', 9093) IPv4]
[2019-09-25T12:15:39+0500] INFO kafka.conn Probing node 1 broker version
[2019-09-25T12:15:39+0500] ERROR kafka.conn Connect attempt to <BrokerConnection node_id=1 host=192.168.3.102:9093 <connecting> [IPv4 ('192.168.3.102', 9093)]> returned error 111. Disconnecting.
[2019-09-25T12:15:39+0500] INFO kafka.conn <BrokerConnection node_id=1 host=192.168.3.102:9093 <connecting> [IPv4 ('192.168.3.102', 9093)]>: Closing connection. ConnectionError: 111 ECONNREFUSED
[2019-09-25T12:15:40+0500] INFO kafka.conn <BrokerConnection node_id=1 host=192.168.3.102:9093 <connecting> [IPv4 ('192.168.3.102', 9093)]>: connecting to 192.168.3.102:9093 [('192.168.3.102', 9093) IPv4]
[2019-09-25T12:15:40+0500] ERROR kafka.conn Connect attempt to <BrokerConnection node_id=1 host=192.168.3.102:9093 <connecting> [IPv4 ('192.168.3.102', 9093)]> returned error 111. Disconnecting.
[2019-09-25T12:15:40+0500] INFO kafka.conn <BrokerConnection node_id=1 host=192.168.3.102:9093 <connecting> [IPv4 ('192.168.3.102', 9093)]>: Closing connection. ConnectionError: 111 ECONNREFUSED
[2019-09-25T12:15:40+0500] INFO Activity URL collector Exception in activity url collector: NoBrokersAvailable

从日志来看,似乎已建立连接,即

<connecting> [IPv4 ('192.168.3.102', 30035)]>: Connection complete.
Bootstrap succeeded: found 1 brokers and 26 topics.

但后来就断线了。

请帮我弄清楚我缺少什么以及如何解决这个问题。谢谢。

您正在引导到端口 30035,并获得初始连接,然后为后续连接返回播发的 9093 端口,而不是 30035。

您需要 NodePort 和广告中的相同,或者至少两者都可以在外部路由。如果您的代码在您的主机

上是 运行,那么您还需要在您的 VM 上进行端口转发

注意:存在用于在 Kubernetes 中设置 Kafka 的 Confluent 或 Strimzi Helm Charts

AWS/DigitalOcean/GCE/Azure 中 Kubernetes 的 Kafka/Zookeeper 简单配置,具有外部访问权限:

https://github.com/StanislavKo/k8s_digitalocean_kafka

您可以通过常规二进制协议从 AWS/DO/GCE 外部连接到 Kafka。连接是 PLAINTEXT 或 SASL_PLAINTEXT (username/password).

Kafka 集群是 StatefulSet,因此您可以轻松扩展集群。

我还会为 Kubernetes 上的 Kafka 推荐 Strimzi。对于外部访问这篇文章救了我https://developers.redhat.com/blog/2019/06/11/accessing-apache-kafka-in-strimzi-part-4-load-balancers/。 我的配置如下:

apiVersion: kafka.strimzi.io/v1beta1
kind: Kafka
metadata:
  name: my-cluster
spec:
  kafka:
    version: 2.4.0
    replicas: 1
    listeners:
      plain: {}
      tls: {}
      external:
        type: loadbalancer
        tls: false
    config:
      offsets.topic.replication.factor: 1
      transaction.state.log.replication.factor: 1
      transaction.state.log.min.isr: 1
      log.message.format.version: "2.4"
    storage:
      type: ephemeral
  zookeeper:
    replicas: 1
    storage:
      type: ephemeral
  entityOperator:
    topicOperator: {}
    userOperator: {}

这用于检索 IP

kubectl get service my-cluster-kafka-external-bootstrap -o=jsonpath='{.status.loadBalancer.ingress[0].ip}{"\n"}'