小知识：「云原生」Apache Livy on k8s 讲解与实战操作

一、概述

Livy是一个提供Rest接口和spark集群交互的服务。它可以提交Spark Job或者Spark一段代码,同步或者异步的返回结果;也提供Sparkcontext的管理,通过Restful接口或RPC客户端库。Livy也简化了与Spark与应用服务的交互,这允许通过web/mobile与Spark的使用交互。

官网：https://livy.incubator.apache.org/GitHub地址：https://github.com/apache/incubator-livy关于Apache Livy更多介绍也可以参考我这篇文章：Spark开源REST服务——Apache Livy（Spark 客户端）

二、开始编排部署

1.部署包准备

这里也提供上面编译好的livy部署包，有需要的小伙伴可以自行下载：

链接：https://pan.baidu.com/s/1pPCbe0lUJ6ji8rvQYsVw9A?pwd=qn7i提取码：qn7i

1）构建镜像

Dockerfile

FROM myharbor.com/bigdata/centos:7.9.2009 RUN rm –f /etc/localtime && ln –sv /usr/share/zoneinfo/Asia/Shanghai /etc/localtime && echo “Asia/Shanghai” > /etc/timezone RUN export LANG=zh_CN.UTF–8 ### install tools RUN yum install –y vim tar wget curl less telnet net–tools lsof RUN groupadd –system –gid=9999 admin && useradd –system -m /home/admin –uid=9999 –gid=admin admin RUN mkdir –p /opt/apache ADD apache–livy–0.8.0–incubating–SNAPSHOT–bin.zip /opt/apache/ ENV LIVY_HOME=/opt/apache/apache–livy RUN ln –s /opt/apache/apache–livy–0.8.0–incubating–SNAPSHOT–bin $LIVY_HOME ADD hadoop–3.3.2.tar.gz /opt/apache/ ENV HADOOP_HOME=/opt/apache/hadoop RUN ln –s /opt/apache/hadoop–3.3.2 $HADOOP_HOME ENV HADOOP_CONFIG_DIR=${HADOOP_HOME}/etc/hadoop ADD spark–3.3.0–bin–hadoop3.tar.gz /opt/apache/ ENV SPARK_HOME=/opt/apache/spark RUN ln –s /opt/apache/spark–3.3.0–bin–hadoop3 $SPARK_HOME ENV PATH=${LIVY_HOME}/bin:${HADOOP_HOME}/bin:${SPARK_HOME}/bin:$PATH RUN chown –R admin:admin /opt/apache WORKDIR $LIVY_HOME ENTRYPOINT ${LIVY_HOME}/bin/livy–server start;tail –f ${LIVY_HOME}/logs/livy–root–server.out

【注意】hadoop包里的core-site.xml，hdfs-site.xml，yarn-site.xml

开始构建镜像

docker build –t myharbor.com/bigdata/livy:0.8.0 . –no-cache ### 参数解释 # –t：指定镜像名称 # . ：当前目录Dockerfile # –f：指定Dockerfile路径 # –no-cache：不缓存 # 推送到harbor docker push myharbor.com/bigdata/livy:0.8.0 2）创建livy chart模板 helm create livy 3）修改yaml编排

livy/values.yaml

replicaCount: 1 image: repository: myharbor.com/bigdata/livy pullPolicy: IfNotPresent # Overrides the image tag whose default is the chart appVersion. tag: “0.8.0” securityContext: runAsUser: 9999 runAsGroup: 9999 privileged: true service: type: NodePort port: 8998 nodePort: 31998

livy/templates/configmap.yaml

apiVersion: v1 kind: ConfigMap metadata: name: {{ include “livy.fullname” . }} labels: {{– include “livy.labels” . | nindent 4 }} data: livy.conf: |- livy.spark.master = yarn livy.spark.deploy–mode = client livy.environment = production livy.impersonation.enabled = true livy.server.csrf_protection.enabled = false livy.server.port = {{ .Values.service.port }} livy.server.session.timeout = 3600000 livy.server.recovery.mode = recovery livy.server.recovery.state–store = filesystem livy.server.recovery.state–store.url = /tmp/livy livy.repl.enable–hive–context = true livy–env.sh: |- export JAVA_HOME=/opt/apache/jdk1.8.0_212 export HADOOP_HOME=/opt/apache/hadoop export HADOOP_CONF_DIR=/opt/apache/hadoop/etc/hadoop export SPARK_HOME=/opt/apache/spark export SPARK_CONF_DIR=/opt/apache/spark/conf export LIVY_LOG_DIR=/opt/apache/livy/logs export LIVY_PID_DIR=/opt/apache/livy/pid–dir export LIVY_SERVER_JAVA_OPTS=“-Xmx512m” spark–blacklist.conf: |- spark.master spark.submit.deployMode # Disallow overriding the location of Spark cached jars. spark.yarn.jar spark.yarn.jars spark.yarn.archive # Dont allow users to override the RSC timeout. livy.rsc.server.idle-timeout

livy/templates/deployment.yaml

apiVersion: apps/v1 kind: Deployment metadata: name: {{ include “livy.fullname” . }} labels: {{– include “livy.labels” . | nindent 4 }} spec: {{– if not .Values.autoscaling.enabled }} replicas: {{ .Values.replicaCount }} {{– end }} selector: matchLabels: {{– include “livy.selectorLabels” . | nindent 6 }} template: metadata: {{– with .Values.podAnnotations }} annotations: {{– toYaml . | nindent 8 }} {{– end }} labels: {{– include “livy.selectorLabels” . | nindent 8 }} spec: {{– with .Values.imagePullSecrets }} imagePullSecrets: {{– toYaml . | nindent 8 }} {{– end }} serviceAccountName: {{ include “livy.serviceAccountName” . }} securityContext: {{– toYaml .Values.podSecurityContext | nindent 8 }} containers: – name: {{ .Chart.Name }} securityContext: {{– toYaml .Values.securityContext | nindent 12 }} image: “{{ .Values.image.repository }}:{{ .Values.image.tag | default .Chart.AppVersion }}” imagePullPolicy: {{ .Values.image.pullPolicy }} ports: – name: http containerPort: 8998 protocol: TCP livenessProbe: httpGet: path: / port: http readinessProbe: httpGet: path: / port: http resources: {{– toYaml .Values.resources | nindent 12 }} {{– with .Values.securityContext }} securityContext: runAsUser: {{ .runAsUser }} runAsGroup: {{ .runAsGroup }} privileged: {{ .privileged }} {{– end }} volumeMounts: – name: {{ .Release.Name }}–livy–conf mountPath: /opt/apache/livy/conf/livy.conf subPath: livy.conf – name: {{ .Release.Name }}–livy–env mountPath: /opt/apache/livy/conf/livy–env.sh subPath: livy–env.sh – name: {{ .Release.Name }}–spark–blacklist–conf mountPath: /opt/apache/livy/conf/spark–blacklist.conf subPath: spark–blacklist.conf {{– with .Values.nodeSelector }} nodeSelector: {{– toYaml . | nindent 8 }} {{– end }} {{– with .Values.affinity }} affinity: {{– toYaml . | nindent 8 }} {{– end }} {{– with .Values.tolerations }} tolerations: {{– toYaml . | nindent 8 }} {{– end }} volumes: – name: {{ .Release.Name }}–livy–conf configMap: name: {{ include “livy.fullname” . }} – name: {{ .Release.Name }}–livy–env configMap: name: {{ include “livy.fullname” . }} – name: {{ .Release.Name }}–spark–blacklist–conf configMap: name: {{ include “livy.fullname” . }} 4）开始部署 helm install livy ./livy –n livy –create-namespace

NOTES

NOTES: 1. Get the application URL by running these commands: export NODE_PORT=$(kubectl get –namespace livy -o jsonpath=”{.spec.ports[0].nodePort}” services livy) export NODE_IP=$(kubectl get nodes –namespace livy -o jsonpath=”{.items[0].status.addresses[0].address}”) echo http://$NODE_IP:$NODE_PORT

查看

kubectl get pods,svc –n livy –owide

web地址：http://192.168.182.110:31998/ui

5）测试验证 curl –s –XPOST –d {“file”:”hdfs://myhdfs/tmp/spark-examples_2.12-3.3.0.jar”,”className”:”org.apache.spark.examples.SparkPi”,”name”:”SparkPi-test”} –H “Content-Type: application/json” http://local–168–182–110:31998/batches|python –m json.tool

6）卸载 helm uninstall livy –n livy

git地址：https://gitee.com/hadoop-bigdata/livy-on-k8s

原文地址：https://www.toutiao.com/article/7163292224204046887/