在Kubernetes中搭建日志系统

乔克的好奇心 2021-09-27

338

【微语】无论你什么时候开始，重要的是开始了就不要停止；无论你什么时候结束，重要的是结束了就不要悔恨。

参考资料：
作者：阳明
博客地址：https://www.qikqiak.com/k8s-book/

Kubernetes 中比较流行的日志收集解决方案是 Elasticsearch、Fluentd 和 Kibana（EFK）技术栈，也是官方现在比较推荐的一种方案。

Elasticsearch 是一个实时的、分布式的可扩展的搜索引擎，允许进行全文、结构化搜索，它通常用于索引和搜索大量日志数据，也可用于搜索许多不同类型的文档。
Kibana 是 Elasticsearch 的一个功能强大的数据可视化 Dashboard，Kibana 允许你通过 web 界面来浏览 Elasticsearch 日志数据。
Fluentd是一个流行的开源数据收集器，我们将在 Kubernetes 集群节点上安装 Fluentd，通过获取容器日志文件、过滤和转换日志数据，然后将数据传递到 Elasticsearch 集群，在该集群中对其进行索引和存储。

正常情况下，上面这种方案就足够我们使用，但是如果集群日志太多，ES不堪重负，我们就需要接入中间件来缓冲数据，对于这些中间件来说kafka和redis无疑是我们的首选方案。我们这里采用了kafka，我们追求一切容器化，所以将这些组件全部都部署在Kubernetes中。

注：

（1）、我们将所有的组件都部署在一个单独的namespace中，我这里是新建了一个kube-ops的namespace；

（2）、集群部署到分布式存储，可选ceph，NFS等，我这里采用的NFS，如果你和我一样使用NFS并且不会搭建，可以参考https://www.qikqiak.com/k8s-book/docs/35.StorageClass.html；

创建Namespace

首先创建一个Namespace，可以使用命令，如下：

kubectl create ns kube-ops

也可以使用YAML清单，如下（efk-ns.yaml）：

apiVersion: v1
kind: Namespace
metadata:
  name: kube-ops

如果使用清单，需要创建清单文件：

kubectl apply -f efk-ns.yaml

部署Elasticsearch

首先我们来部署Elasticsearch集群。

开始部署3个节点的ElasticSearch。其中关键点是应该设置discover.zen.minimummasternodes=N/2+1，其中N是 Elasticsearch 集群中符合主节点的节点数，比如我们这里3个节点，意味着N应该设置为2。这样，如果一个节点暂时与集群断开连接，则另外两个节点可以选择一个新的主节点，并且集群可以在最后一个节点尝试重新加入时继续运行，在扩展 Elasticsearch 集群时，一定要记住这个参数。

（1）、创建 elasticsearch无头服务（elasticsearch-svc.yaml）

apiVersion: v1
kind: Service
metadata:
  name: elasticsearch
  namespace: kube-ops
  labels:
    app: elasticsearch
spec:
  selector:
    app: elasticsearch
  clusterIP: None
  ports:
  - name: rest
    port: 9200
  - name: inter-node
    port: 9300

定义为无头服务，是因为我们后面真正部署elasticsearch的pod是通过statefulSet部署的，到时候将其进行关联，另外9200是REST API端口，9300是集群间通信端口。

然后我们创建这个资源对象。

# kubectl apply -f elasticsearch-svc.yaml
service/elasticsearch created
# kubectl get svc -n kube-ops
NAME            TYPE        CLUSTER-IP      EXTERNAL-IP   PORT(S)             AGE
elasticsearch   ClusterIP   None            <none>        9200/TCP,9300/TCP   9s

（2）、用StatefulSet部署Elasticsearch，配置清单如下（elasticsearch-elasticsearch.yaml ）：

apiVersion: apps/v1
kind: StatefulSet
metadata:
  name: es-cluster
  namespace: kube-ops
spec:
  serviceName: elasticsearch
  replicas: 3
  selector:
    matchLabels:
      app: elasticsearch
  template:
    metadata:
      labels:
        app: elasticsearch
    spec:
      containers:
      - name: elasticsearch
        image: docker.elastic.co/elasticsearch/elasticsearch-oss:6.4.3
        resources:
            limits:
              cpu: 1000m
            requests:
              cpu: 1000m
        ports:
        - containerPort: 9200
          name: rest
          protocol: TCP
        - containerPort: 9300
          name: inter-node
          protocol: TCP
        volumeMounts:
        - name: data
          mountPath: /usr/share/elasticsearch/data
        env:
          - name: cluster.name
            value: k8s-logs
          - name: node.name
            valueFrom:
              fieldRef:
                fieldPath: metadata.name
          - name: discovery.zen.ping.unicast.hosts
            value: "es-cluster-0.elasticsearch,es-cluster-1.elasticsearch,es-cluster-2.elasticsearch"
          - name: discovery.zen.minimum_master_nodes
            value: "2"
          - name: ES_JAVA_OPTS
            value: "-Xms512m -Xmx512m"
      initContainers:
      - name: fix-permissions
        image: busybox
        command: ["sh", "-c", "chown -R 1000:1000 usr/share/elasticsearch/data"]
        securityContext:
          privileged: true
        volumeMounts:
        - name: data
          mountPath: /usr/share/elasticsearch/data
      - name: increase-vm-max-map
        image: busybox
        command: ["sysctl", "-w", "vm.max_map_count=262144"]
        securityContext:
          privileged: true
      - name: increase-fd-ulimit
        image: busybox
        command: ["sh", "-c", "ulimit -n 65536"]
        securityContext:
          privileged: true
  volumeClaimTemplates:
  - metadata:
      name: data
      labels:
        app: elasticsearch
      annotations:
        volume.beta.kubernetes.io/storage-class: es-data-db
    spec:
      accessModes: [ "ReadWriteOnce" ]
      storageClassName: es-data-db
      resources:
        requests:
          storage: 10Gi

配置清单说明：

上面Pod中定义了两种类型的container，普通的container和initContainer。其中在initContainer中它有3个container，它们会在所有容器启动前运行。

名为fix-permissions的container的作用是将 Elasticsearch 数据目录的用户和组更改为1000:1000（Elasticsearch 用户的 UID）。因为默认情况下，Kubernetes 用 root 用户挂载数据目录，这会使得 Elasticsearch 无法读取该数据目录。
名为 increase-vm-max-map 的容器用来增加操作系统对mmap计数的限制，默认情况下该值可能太低，导致内存不足的错误
名为increase-fd-ulimit的容器用来执行ulimit命令增加打开文件描述符的最大数量

在普通container中，我们定义了名为elasticsearch的container，然后暴露了9200和9300两个端口，注意名称要和上面定义的 Service 保持一致。然后通过 volumeMount 声明了数据持久化目录，下面我们再来定义 VolumeClaims。最后就是我们在容器中设置的一些环境变量了：

cluster.name：Elasticsearch 集群的名称，我们这里命名成 k8s-logs；
node.name：节点的名称，通过metadata.name来获取。这将解析为 es-cluster-[0,1,2]，取决于节点的指定顺序；
discovery.zen.ping.unicast.hosts：此字段用于设置在 Elasticsearch 集群中节点相互连接的发现方法。我们使用 unicastdiscovery 方式，它为我们的集群指定了一个静态主机列表。由于我们之前配置的无头服务，我们的 Pod 具有唯一的 DNS 域es-cluster-[0,1,2].elasticsearch.logging.svc.cluster.local，因此我们相应地设置此变量。由于都在同一个 namespace 下面，所以我们可以将其缩短为es-cluster-[0,1,2].elasticsearch；
discovery.zen.minimummasternodes：我们将其设置为(N/2) + 1，N是我们的群集中符合主节点的节点的数量。我们有3个 Elasticsearch 节点，因此我们将此值设置为2（向下舍入到最接近的整数）；
ESJAVAOPTS：这里我们设置为-Xms512m -Xmx512m，告诉JVM使用512 MB的最小和最大堆。您应该根据群集的资源可用性和需求调整这些参数；

（3）、定义一个StorageClass（elasticsearch-storage.yaml）

apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: es-data-db
provisioner: rookieops/nfs

注意：由于我们这里采用的是NFS来存储，所以上面的provisioner需要和我们nfs-client-provisoner中保持一致。

然后我们创建资源：

# kubectl apply -f elasticsearch-storage.yaml
# kubectl apply -f elasticsearch-elasticsearch.yaml
# kubectl get pod -n kube-ops
NAME                             READY   STATUS    RESTARTS   AGE
dingtalk-hook-8497494dc6-s6qkh   1/1     Running   0          16m
es-cluster-0                     1/1     Running   0          10m
es-cluster-1                     1/1     Running   0          10m
es-cluster-2                     1/1     Running   0          9m20s
# kubectl get pvc -n kube-ops
NAME                STATUS   VOLUME                                     CAPACITY   ACCESS MODES   STORAGECLASS   AGE
data-es-cluster-0   Bound    pvc-9f15c0f8-60a8-485d-b650-91fb8f5f8076   10Gi       RWO            es-data-db     18m
data-es-cluster-1   Bound    pvc-503828ec-d98e-4e94-9f00-eaf6c05f3afd   10Gi       RWO            es-data-db     11m
data-es-cluster-2   Bound    pvc-3d2eb82e-396a-4eb0-bb4e-2dd4fba8600e   10Gi       RWO            es-data-db     10m
# kubectl get svc -n kube-ops
NAME            TYPE        CLUSTER-IP     EXTERNAL-IP   PORT(S)             AGE
dingtalk-hook   ClusterIP   10.68.122.48   <none>        5000/TCP            18m
elasticsearch   ClusterIP   None           <none>        9200/TCP,9300/TCP   19m

测试：

# kubectl port-forward es-cluster-0 9200:9200 --namespace=kube-ops
Forwarding from 127.0.0.1:9200 -> 9200
Forwarding from [::1]:9200 -> 9200
Handling connection for 9200

如果看到如下结果，就表示服务正常：

# curl http://localhost:9200/_cluster/state?pretty
{
  "cluster_name" : "k8s-logs",
  "compressed_size_in_bytes" : 337,
  "cluster_uuid" : "nzc4y-eDSuSaYU1TigFAWw",
  "version" : 3,
  "state_uuid" : "6Mvd-WTPT0e7WMJV23Vdiw",
  "master_node" : "KRyMrbS0RXSfRkpS0ZaarQ",
  "blocks" : { },
  "nodes" : {
    "XGP4TrkrQ8KNMpH3pQlaEQ" : {
      "name" : "es-cluster-2",
      "ephemeral_id" : "f-R_IyfoSYGhY27FmA41Tg",
      "transport_address" : "172.20.1.104:9300",
      "attributes" : { }
    },
    "KRyMrbS0RXSfRkpS0ZaarQ" : {
      "name" : "es-cluster-0",
      "ephemeral_id" : "FpTnJTR8S3ysmoZlPPDnSg",
      "transport_address" : "172.20.1.102:9300",
      "attributes" : { }
    },
    "Xzjk2n3xQUutvbwx2h7f4g" : {
      "name" : "es-cluster-1",
      "ephemeral_id" : "FKjRuegwToe6Fz8vgPmSNw",
      "transport_address" : "172.20.1.103:9300",
      "attributes" : { }
    }
  },
  "metadata" : {
    "cluster_uuid" : "nzc4y-eDSuSaYU1TigFAWw",
    "templates" : { },
    "indices" : { },
    "index-graveyard" : {
      "tombstones" : [ ]
    }
  },
  "routing_table" : {
    "indices" : { }
  },
  "routing_nodes" : {
    "unassigned" : [ ],
    "nodes" : {
      "KRyMrbS0RXSfRkpS0ZaarQ" : [ ],
      "XGP4TrkrQ8KNMpH3pQlaEQ" : [ ],
      "Xzjk2n3xQUutvbwx2h7f4g" : [ ]
    }
  },
  "snapshots" : {
    "snapshots" : [ ]
  },
  "restore" : {
    "snapshots" : [ ]
  },
  "snapshot_deletions" : {
    "snapshot_deletions" : [ ]
  }
}

到此，Elasticsearch部署完成。

部署kibana

对于kibana，它只是一个展示工具，所以我们用Deployment部署即可。

（1）、定义kibana service的配置清单（kibana-svc.yaml）

apiVersion: v1
kind: Service
metadata:
  name: kibana
  namespace: kube-ops
  labels:
    app: kibana
spec:
  ports:
  - port: 5601
  type: NodePort
  selector:
    app: kibana

我们这里配置的Service是采用NodePort类型，当然也可以采用ingress，推荐使用ingress。

（2）、定义kibana Deployment配置清单（kibana-deploy.yaml）

apiVersion: apps/v1
kind: Deployment
metadata:
  name: kibana
  namespace: kube-ops
  labels:
    app: kibana
spec:
  selector:
    matchLabels:
      app: kibana
  template:
    metadata:
      labels:
        app: kibana
    spec:
      containers:
      - name: kibana
        image: docker.elastic.co/kibana/kibana-oss:6.4.3
        resources:
          limits:
            cpu: 1000m
          requests:
            cpu: 100m
        env:
          - name: ELASTICSEARCH_URL
            value: http://elasticsearch:9200
        ports:
        - containerPort: 5601

创建配置清单：

# kubectl apply -f kibana.yaml
service/kibana created
deployment.apps/kibana created

# kubectl get svc -n kube-ops
NAME            TYPE        CLUSTER-IP     EXTERNAL-IP   PORT(S)             AGE
dingtalk-hook   ClusterIP   10.68.122.48   <none>        5000/TCP            47m
elasticsearch   ClusterIP   None           <none>        9200/TCP,9300/TCP   48m
kibana          NodePort    10.68.221.60   <none>        5601:26575/TCP      7m29s
[root@ecs-5704-0003 storage]# kubectl get pod -n kube-ops
NAME                             READY   STATUS    RESTARTS   AGE
dingtalk-hook-8497494dc6-s6qkh   1/1     Running   0          47m
es-cluster-0                     1/1     Running   0          41m
es-cluster-1                     1/1     Running   0          41m
es-cluster-2                     1/1     Running   0          40m
kibana-7fc9f8c964-68xbh          1/1     Running   0          7m41s

如果看到以下界面，以表示你的kibana部署完成。

部署kafka

Apache Kafka是一个分布式发布 - 订阅消息系统和一个强大的队列，可以处理大量的数据，并使您能够将消息从一个端点传递到另一个端点。Kafka适合离线和在线消息消费。Kafka消息保留在磁盘上，并在群集内复制以防止数据丢失。

以下是Kafka的几个好处：

可靠性 - Kafka是分布式，分区，复制和容错的。
可扩展性 - Kafka消息传递系统轻松缩放，无需停机。
耐用性 - Kafka使用分布式提交日志，这意味着消息会尽可能快地保留在磁盘上，因此它是持久的。
性能 - Kafka对于发布和订阅消息都具有高吞吐量。即使存储了许多TB的消息，它也保持稳定的性能。

Kafka的一个关键依赖是Zookeeper，它是一个分布式配置和同步服务， Zookeeper是Kafka代理和消费者之间的协调接口， Kafka服务器通过Zookeeper集群共享信息。Kafka在Zookeeper中存储基本元数据，例如关于主题，代理，消费者偏移(队列读取器)等的信息。

由于所有关键信息存储在Zookeeper中，并且它通常在其整体上复制此数据，因此Kafka代理/ Zookeeper的故障不会影响Kafka集群的状态，另外Kafka代理之间的领导者选举也通过使用Zookeeper在领导者失败的情况下完成的。

部署zookeeper

（1）、定义ZK的storageClass（zookeeper-storage.yaml）

apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: zk-data-db
provisioner: rookieops/nfs

（2）、定义ZK的headless service（zookeeper-svc.yaml）

apiVersion: v1
kind: Service
metadata:
  name: zk-svc
  namespace: kube-ops
  labels:
    app: zk-svc
spec:
  ports:
  - port: 2888
    name: server
  - port: 3888
    name: leader-election
  clusterIP: None
  selector:
    app: zk

（3）、定义ZK的configMap（zookeeper-config.yaml）

apiVersion: v1
kind: ConfigMap
metadata:
  name: zk-cm
  namespace: kube-ops
data:
  jvm.heap: "1G"
  tick: "2000"
  init: "10"
  sync: "5"
  client.cnxns: "60"
  snap.retain: "3"
  purge.interval: "0"

（4）、定义ZK的PodDisruptionBudget（zookeeper-pdb.yaml）

apiVersion: policy/v1beta1
kind: PodDisruptionBudget
metadata:
  name: zk-pdb
  namespace: kube-ops
spec:
  selector:
    matchLabels:
      app: zk
  minAvailable: 2

说明：PodDisruptionBudget的作用就是为了保证业务不中断或者业务SLA不降级。通过PodDisruptionBudget控制器可以设置应用POD集群处于运行状态最低个数，也可以设置应用POD集群处于运行状态的最低百分比，这样可以保证在主动销毁应用POD的时候，不会一次性销毁太多的应用POD，从而保证业务不中断或业务SLA不降级。

（5）、定义ZK的statefulSet（zookeeper-statefulset.yaml）

apiVersion: apps/v1beta1
kind: StatefulSet
metadata:
  name: zk
  namespace: kube-ops
spec:
  serviceName: zk-svc
  replicas: 3
  template:
    metadata:
      labels:
        app: zk
    spec:
      containers:
      - name: k8szk
        imagePullPolicy: Always
        image: registry.cn-hangzhou.aliyuncs.com/rookieops/zookeeper:3.4.10
        resources:
          requests:
            memory: "2Gi"
            cpu: "500m"
        ports:
        - containerPort: 2181
          name: client
        - containerPort: 2888
          name: server
        - containerPort: 3888
          name: leader-election
        env:
        - name : ZK_REPLICAS
          value: "3"
        - name : ZK_HEAP_SIZE
          valueFrom:
            configMapKeyRef:
                name: zk-cm
                key: jvm.heap
        - name : ZK_TICK_TIME
          valueFrom:
            configMapKeyRef:
                name: zk-cm
                key: tick
        - name : ZK_INIT_LIMIT
          valueFrom:
            configMapKeyRef:
                name: zk-cm
                key: init
        - name : ZK_SYNC_LIMIT
          valueFrom:
            configMapKeyRef:
                name: zk-cm
                key: tick
        - name : ZK_MAX_CLIENT_CNXNS
          valueFrom:
            configMapKeyRef:
                name: zk-cm
                key: client.cnxns
        - name: ZK_SNAP_RETAIN_COUNT
          valueFrom:
            configMapKeyRef:
                name: zk-cm
                key: snap.retain
        - name: ZK_PURGE_INTERVAL
          valueFrom:
            configMapKeyRef:
                name: zk-cm
                key: purge.interval
        - name: ZK_CLIENT_PORT
          value: "2181"
        - name: ZK_SERVER_PORT
          value: "2888"
        - name: ZK_ELECTION_PORT
          value: "3888"
        command:
        - sh
        - -c
        - zkGenConfig.sh && zkServer.sh start-foreground
        readinessProbe:
          exec:
            command:
            - "zkOk.sh"
          initialDelaySeconds: 10
          timeoutSeconds: 5
        livenessProbe:
          exec:
            command:
            - "zkOk.sh"
          initialDelaySeconds: 10
          timeoutSeconds: 5
        volumeMounts:
        - name: datadir
          mountPath: /var/lib/zookeeper
  volumeClaimTemplates:
  - metadata:
      name: datadir
    spec:
      accessModes: ["ReadWriteOnce"]
      storageClassName: zk-data-db
      resources:
        requests:
          storage: 1Gi

然后创建配置清单：

# kubectl apply -f zookeeper-storage.yaml
# kubectl apply -f zookeeper-svc.yaml
# kubectl apply -f zookeeper-config.yaml
# kubectl apply -f zookeeper-pdb.yaml
# kubectl apply -f zookeeper-statefulset.yaml
# kubectl get pod -n kube-ops
NAME                             READY   STATUS    RESTARTS   AGE
zk-0                             1/1     Running   0          12m
zk-1                             1/1     Running   0          12m
zk-2                             1/1     Running   0          11m

然后查看集群状态：

# for i in 0 1 2; do kubectl exec -n kube-ops zk-$i zkServer.sh status; done
ZooKeeper JMX enabled by default
Using config: /usr/bin/../etc/zookeeper/zoo.cfg
Mode: follower
ZooKeeper JMX enabled by default
Using config: /usr/bin/../etc/zookeeper/zoo.cfg
Mode: follower
ZooKeeper JMX enabled by default
Using config: /usr/bin/../etc/zookeeper/zoo.cfg
Mode: leader

部署kafka

（1）、制作镜像，Dokcerfile如下：

kafka下载地址：wget https://www-us.apache.org/dist/kafka/2.2.0/kafka_2.11-2.2.0.tgz

FROM centos:centos7
LABEL "auth"="rookieops" \
      "mail"="rookieops@163.com"
ENV TIME_ZONE Asia/Shanghai

# install JAVA
ADD jdk-8u131-linux-x64.tar.gz /opt/
ENV JAVA_HOME /opt/jdk1.8.0_131
ENV PATH ${JAVA_HOME}/bin:${PATH}

# install kafka
ADD kafka_2.11-2.3.1.tgz /opt/
RUN mv /opt/kafka_2.11-2.3.1 /opt/kafka
WORKDIR /opt/kafka
EXPOSE 9092 
CMD ["./bin/kafka-server-start.sh", "config/server.properties"]

然后docker build，docker push到镜像仓库（操作略）。

（2）、定义kafka的storageClass（kafka-storage.yaml ）

apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: kafka-data-db
provisioner: rookieops/nfs

（3）、定义kafka headless Service（kafka-svc.yaml）

apiVersion: v1
kind: Service
metadata:
  name: kafka-svc
  namespace: kube-ops
  labels:
    app: kafka
spec:
  selector:
    app: kafka
  clusterIP: None
  ports:
  - name: server
    port: 9092

（4）、定义kafka的configMap（kafka-config.yaml）

apiVersion: v1
kind: ConfigMap
metadata:
  name: kafka-config
  namespace: kube-ops
data:
  server.properties: |
    broker.id=${HOSTNAME##*-}
    listeners=PLAINTEXT://:9092
    num.network.threads=3
    num.io.threads=8
    socket.send.buffer.bytes=102400
    socket.receive.buffer.bytes=102400
    socket.request.max.bytes=104857600
    log.dirs=/data/kafka/logs
    num.partitions=1
    num.recovery.threads.per.data.dir=1
    offsets.topic.replication.factor=1
    transaction.state.log.replication.factor=1
    transaction.state.log.min.isr=1
    log.retention.hours=168
    log.segment.bytes=1073741824
    log.retention.check.interval.ms=300000
    zookeeper.connect=zk-0.zk-svc.kube-ops.svc.cluster.local:2181,zk-1.zk-svc.kube-ops.svc.cluster.local:2181,zk-2.zk-svc.kube-ops.svc.cluster.local:2181
    zookeeper.connection.timeout.ms=6000
    group.initial.rebalance.delay.ms=0

（5）、定义kafka的statefuleSet配置清单（kafka.yaml）

apiVersion: apps/v1
kind: StatefulSet
metadata:
  name: kafka
  namespace: kube-ops
spec:
  serviceName: kafka-svc
  replicas: 3
  selector:
    matchLabels:
      app: kafka
  template:
    metadata:
      labels:
        app: kafka
    spec:
      affinity:
        podAffinity:
          preferredDuringSchedulingIgnoredDuringExecution:
             - weight: 1
               podAffinityTerm:
                 labelSelector:
                    matchExpressions:
                      - key: "app"
                        operator: In
                        values: 
                        - zk
                 topologyKey: "kubernetes.io/hostname"
      terminationGracePeriodSeconds: 300
      containers:
      - name: kafka
        image: registry.cn-hangzhou.aliyuncs.com/rookieops/kafka:2.3.1-beta
        imagePullPolicy: Always
        resources:
          requests:
            cpu: 500m
            memory: 1Gi
          limits:
            cpu: 500m
            memory: 1Gi
        command:
        - "/bin/sh"
        - "-c"
        - "./bin/kafka-server-start.sh config/server.properties --override broker.id=${HOSTNAME##*-}"
        ports:
        - name: server
          containerPort: 9092
        volumeMounts:
        - name: config
          mountPath: /opt/kafka/config/server.properties
          subPath: server.properties
        - name: data
          mountPath: /data/kafka/logs
      volumes:
      - name: config
        configMap:
          name: kafka-config
  volumeClaimTemplates:
  - metadata:
      name: data
    spec:
      accessModes: [ "ReadWriteOnce" ]
      storageClassName: kafka-data-db
      resources:
        requests:
          storage: 10Gi

创建配置清单：

# kubectl apply -f kafka-storage.yaml
# kubectl apply -f kafka-svc.yaml
# kubectl apply -f kafka-config.yaml
# kubectl apply -f kafka.yaml
# kubectl get pod -n kube-ops
NAME                             READY   STATUS    RESTARTS   AGE
kafka-0                          1/1     Running   0          13m
kafka-1                          1/1     Running   0          13m
kafka-2                          1/1     Running   0          10m
zk-0                             1/1     Running   0          77m
zk-1                             1/1     Running   0          77m
zk-2                             1/1     Running   0          76m

测试：

（1）、进入一个container，创建topic，并开启consumer等待producer生产数据

# kubectl exec -it -n kube-ops kafka-0 -- bin/bash
$ cd /opt/kafka
$ ./bin/kafka-topics.sh --create --topic test --zookeeper zk-0.zk-svc.kube-ops.svc.cluster.local:2181,zk-1.zk-svc.kube-ops.svc.cluster.local:2181,zk-2.zk-svc.kube-ops.svc.cluster.local:2181 --partitions 3 --replication-factor 2
Created topic "test".
# 消费
$ ./bin/kafka-console-consumer.sh --topic test --bootstrap-server localhost:9092

（2）、再进入另一个container做producer：

# kubectl exec -it -n kube-ops kafka-1 -- bin/bash
$ cd /opt/kafka
$ ./bin/kafka-console-producer.sh --topic test --broker-list localhost:9092
hello
nihao

可以看到consumer上会产生消费信息：

$ ./bin/kafka-console-consumer.sh --topic test --bootstrap-server localhost:9092
hello
nihao

至此，kafka集群搭建完成。

部署Logstash

在这里部署Logstash的作用是读取kafka中的信息，然后传递给我们的后端存储ES，为了简化，我这里直接使用Deployment部署了。

（1）、定义configMap配置清单（logstash-config.yaml）

apiVersion: v1
kind: ConfigMap
metadata:
  name: logstash-k8s-config
  namespace: kube-ops
data:
  containers.conf: |
    input {
      kafka {
        codec => "json"
        topics => ["test"]
        bootstrap_servers => ["kafka-0.kafka-svc.kube-ops:9092, kafka-1.kafka-svc.kube-ops:9092, kafka-2.kafka-svc.kube-ops:9092"]
        group_id => "logstash-g1"
      }
    }
    output {
      elasticsearch {
        hosts => ["es-cluster-0.elasticsearch.kube-ops:9200", "es-cluster-1.elasticsearch.kube-ops:9200", "es-cluster-2.elasticsearch.kube-ops:9200"]
        index => "logstash-%{+YYYY.MM.dd}"
      }
    }

（2）、定义Deployment配置清单（logstash.yaml）

kind: Deployment
metadata:
  name: logstash
  namespace: kube-ops
spec:
  replicas: 1
  selector:
    matchLabels:
      app: logstash
  template:
    metadata:
      labels:
        app: logstash
    spec:
      containers:
      - name: logstash
        image: registry.cn-hangzhou.aliyuncs.com/rookieops/logstash-kubernetes:7.1.1
        volumeMounts:
        - name: config
          mountPath: /opt/logstash/config/containers.conf
          subPath: containers.conf
        command:
        - "/bin/sh"
        - "-c"
        - "/opt/logstash/bin/logstash -f opt/logstash/config/containers.conf"
      volumes:
      - name: config
        configMap:
          name: logstash-k8s-config

然后生成配置：

# kubectl apply -f logstash-config.yaml
# kubectl apply -f logstash.yaml

然后观察状态，查看日志：

# kubectl get pod -n kube-ops
NAME                             READY   STATUS    RESTARTS   AGE
dingtalk-hook-856c5dbbc9-srcm6   1/1     Running   0          3d20h
es-cluster-0                     1/1     Running   0          22m
es-cluster-1                     1/1     Running   0          22m
es-cluster-2                     1/1     Running   0          22m
kafka-0                          1/1     Running   0          3h6m
kafka-1                          1/1     Running   0          3h6m
kafka-2                          1/1     Running   0          3h6m
kibana-7fc9f8c964-dqr68          1/1     Running   0          5d2h
logstash-678c945764-lkl2n        1/1     Running   0          10m
zk-0                             1/1     Running   0          3d21h
zk-1                             1/1     Running   0          3d21h
zk-2                             1/1     Running   0          3d21h

部署Fluentd

Fluentd 是一个高效的日志聚合器，是用 Ruby 编写的，并且可以很好地扩展。对于大部分企业来说，Fluentd 足够高效并且消耗的资源相对较少，另外一个工具Fluent-bit更轻量级，占用资源更少，但是插件相对 Fluentd 来说不够丰富，所以整体来说，Fluentd 更加成熟，使用更加广泛，所以我们这里也同样使用 Fluentd 来作为日志收集工具。

（1）、安装fluent-plugin-kafka插件

我这里的安装步骤是先起一个fluentd容器，然后安装插件，最后commit一下，再推送到仓库。具体步骤如下：

a、先用docker起一个容器

# docker run -it registry.cn-hangzhou.aliyuncs.com/rookieops/fluentd-elasticsearch:v2.0.4 bin/bash
$ gem install fluent-plugin-kafka --no-document

b、退出容器，重新commit 一下：

# docker commit c29b250d8df9 registry.cn-hangzhou.aliyuncs.com/rookieops/fluentd-elasticsearch:v2.0.4

c、将安装了插件的镜像推向仓库：

# docker push registry.cn-hangzhou.aliyuncs.com/rookieops/fluentd-elasticsearch:v2.0.4

（2）、定义Fluentd的configMap（fluentd-config.yaml）

kind: ConfigMap
apiVersion: v1
metadata:
  name: fluentd-config
  namespace: kube-ops
  labels:
    addonmanager.kubernetes.io/mode: Reconcile
data:
  system.conf: |-
    <system>
      root_dir /tmp/fluentd-buffers/
    </system>
  containers.input.conf: |-
    <source>
      @id fluentd-containers.log
      @type tail
      path /var/log/containers/*.log
      pos_file /var/log/es-containers.log.pos
      time_format %Y-%m-%dT%H:%M:%S.%NZ
      localtime
      tag raw.kubernetes.*
      format json
      read_from_head true
    </source>
    # Detect exceptions in the log output and forward them as one log entry.
    <match raw.kubernetes.**>
      @id raw.kubernetes
      @type detect_exceptions
      remove_tag_prefix raw
      message log
      stream stream
      multiline_flush_interval 5
      max_bytes 500000
      max_lines 1000
    </match>
  system.input.conf: |-
    # Logs from systemd-journal for interesting services.
    <source>
      @id journald-docker
      @type systemd
      filters [{ "_SYSTEMD_UNIT": "docker.service" }]
      <storage>
        @type local
        persistent true
      </storage>
      read_from_head true
      tag docker
    </source>
    <source>
      @id journald-kubelet
      @type systemd
      filters [{ "_SYSTEMD_UNIT": "kubelet.service" }]
      <storage>
        @type local
        persistent true
      </storage>
      read_from_head true
      tag kubelet
    </source>
  forward.input.conf: |-
    # Takes the messages sent over TCP
    <source>
      @type forward
    </source>
  output.conf: |-
    # Enriches records with Kubernetes metadata
    <filter kubernetes.**>
      @type kubernetes_metadata
    </filter>
    <match **>
      @id kafka
      @type kafka2
      @log_level info
      include_tag_key true
      brokers kafka-0.kafka-svc.kube-ops:9092,kafka-1.kafka-svc.kube-ops:9092,kafka-2.kafka-svc.kube-ops:9092
      logstash_format true
      request_timeout    30s
      <buffer>
        @type file
        path /var/log/fluentd-buffers/kubernetes.system.buffer
        flush_mode interval
        retry_type exponential_backoff
        flush_thread_count 2
        flush_interval 5s
        retry_forever
        retry_max_interval 30
        chunk_limit_size 2M
        queue_limit_length 8
        overflow_action block
      </buffer>
      # data type settings
      <format>
        @type json
      </format>
      # topic settings
      topic_key topic
      default_topic test
      # producer settings
      required_acks -1
      compression_codec gzip
    </match>

（3）、定义DeamonSet配置清单（fluentd-daemonset.yaml）

apiVersion: v1
kind: ServiceAccount
metadata:
  name: fluentd-es
  namespace: kube-ops
  labels:
    k8s-app: fluentd-es
    kubernetes.io/cluster-service: "true"
    addonmanager.kubernetes.io/mode: Reconcile
---
kind: ClusterRole
apiVersion: rbac.authorization.k8s.io/v1
metadata:
  name: fluentd-es
  labels:
    k8s-app: fluentd-es
    kubernetes.io/cluster-service: "true"
    addonmanager.kubernetes.io/mode: Reconcile
rules:
- apiGroups:
  - ""
  resources:
  - "namespaces"
  - "pods"
  verbs:
  - "get"
  - "watch"
  - "list"
---
kind: ClusterRoleBinding
apiVersion: rbac.authorization.k8s.io/v1
metadata:
  name: fluentd-es
  labels:
    k8s-app: fluentd-es
    kubernetes.io/cluster-service: "true"
    addonmanager.kubernetes.io/mode: Reconcile
subjects:
- kind: ServiceAccount
  name: fluentd-es
  namespace: kube-ops
  apiGroup: ""
roleRef:
  kind: ClusterRole
  name: fluentd-es
  apiGroup: ""
---
apiVersion: apps/v1
kind: DaemonSet
metadata:
  name: fluentd-es
  namespace: kube-ops
  labels:
    k8s-app: fluentd-es
    version: v2.0.4
    kubernetes.io/cluster-service: "true"
    addonmanager.kubernetes.io/mode: Reconcile
spec:
  selector:
    matchLabels:
      k8s-app: fluentd-es
      version: v2.0.4
  template:
    metadata:
      labels:
        k8s-app: fluentd-es
        kubernetes.io/cluster-service: "true"
        version: v2.0.4
      # This annotation ensures that fluentd does not get evicted if the node
      # supports critical pod annotation based priority scheme.
      # Note that this does not guarantee admission on the nodes (#40573).
      annotations:
        scheduler.alpha.kubernetes.io/critical-pod: ''
    spec:
      serviceAccountName: fluentd-es
      containers:
      - name: fluentd-es
        image: registry.cn-hangzhou.aliyuncs.com/rookieops/fluentd-elasticsearch:v2.0.4 
        command:
        - "/bin/sh"
        - "-c"
        - "/run.sh $FLUENTD_ARGS"
        env:
        - name: FLUENTD_ARGS
          value: --no-supervisor -q
        resources:
          limits:
            memory: 500Mi
          requests:
            cpu: 100m
            memory: 200Mi
        volumeMounts:
        - name: varlog
          mountPath: /var/log
        - name: varlibdockercontainers
          mountPath: /var/lib/docker/containers
          readOnly: true
        - name: config-volume
          mountPath: /etc/fluent/config.d
      nodeSelector:
        beta.kubernetes.io/fluentd-ds-ready: "true"
      tolerations:
      - key: node-role.kubernetes.io/master
        operator: Exists
        effect: NoSchedule
      terminationGracePeriodSeconds: 30
      volumes:
      - name: varlog
        hostPath:
          path: /var/log
      - name: varlibdockercontainers
        hostPath:
          path: /var/lib/docker/containers
      - name: config-volume
        configMap:
          name: fluentd-config

创建配置清单：

# kubectl apply -f fluentd-daemonset.yaml
# kubectl apply -f fluentd-config.yaml
# kubectl get pod -n kube-ops
NAME                             READY   STATUS    RESTARTS   AGE
dingtalk-hook-856c5dbbc9-srcm6   1/1     Running   0          3d21h
es-cluster-0                     1/1     Running   0          112m
es-cluster-1                     1/1     Running   0          112m
es-cluster-2                     1/1     Running   0          112m
fluentd-es-jvhqv                 1/1     Running   0          4h29m
fluentd-es-s7v6m                 1/1     Running   0          4h29m
kafka-0                          1/1     Running   0          4h36m
kafka-1                          1/1     Running   0          4h36m
kafka-2                          1/1     Running   0          4h36m
kibana-7fc9f8c964-dqr68          1/1     Running   0          5d4h
logstash-678c945764-lkl2n        1/1     Running   0          100m
zk-0                             1/1     Running   0          3d23h
zk-1                             1/1     Running   0          3d23h
zk-2                             1/1     Running   0          3d23h