Skip to content

zeppelin.md

Latest commit

 

History

History
82 lines (71 loc) · 2.64 KB

zeppelin.md

File metadata and controls

82 lines (71 loc) · 2.64 KB

Zeppelin Install

  • Zeppelin

  • Downloads

    • version 0.8.1 download & 압축해제 & soft link
    $ wget http://mirror.apache-kr.org/zeppelin/zeppelin-0.8.1/zeppelin-0.8.1-bin-all.tgz
$ tar -zxf zeppelin-0.8.1-bin-all.tgz
$ ln -sf zeppelin-0.8.1-bin-all zeppelin

  • Configure

    • zeppelin-site.xml
      • zeppelin.server.port : 서버 포트 (default : 8080)
    • .bashrc
      • JAVA_HOME, ZEPPELIN_HOME 추가
    • shiro.ini
      • 로그인 관리
      [users]
admin = password!, admin
user1 = password_user1, role1, role2
user2 = password_user2, role2

  • Nginx

    • proxy 사용시 유의사항
      • 한서버에 여러 web이 올라가다보니 domain 이름별로 proxy pass를 사용함.
      • zeppelin도 동일 방법으로 처리시 /ws에서 문제 발생.
      • /선언 위에 /ws에도 따로 선언 필요!
      server {
    listen       80;
    server_name  zeppelin.test.com;

    access_log  /log/nginx/zeppelin.test.com.access.log  main;
    error_log   /log/nginx/zeppelin.test.com.error.log;

    location /ws {  # For websocket support
    proxy_pass http://127.0.0.1:8282/ws;
    proxy_http_version 1.1;
    proxy_set_header Upgrade websocket;
    proxy_set_header Connection upgrade;
    proxy_read_timeout 86400;
    }

    location / {
        proxy_redirect     off;
        proxy_set_header   Host             $host;
        proxy_set_header   X-Real-IP        $remote_addr;
        proxy_set_header   X-Forwarded-For  $proxy_add_x_forwarded_for;
        proxy_buffers      32 4k;

        # Keepalive
        #proxy_http_version 1.1;
        #proxy_set_header Connection "";
        #proxy_pass http://keepalive_8082;

        proxy_pass http://127.0.0.1:8282;
    }
}

  • 실행

    • ./bin/zeppelin-daemon.sh start

  • Hadoop, Spark 셋팅

    • zeppelin-env.sh
      • export MASTER=yarn-client
      • export SPARK_HOME=스파크홈경로!
      • export HADOOP_CONF_DIR=하둡설정경로!
      • export HADOOP_HOME=하둡홈경로!
      • export HADOOP_LIB_DIR=하둡LIB경로!
      • 그 외 spark setting하고싶으면 해도됨!
        • export SPARK_DRIVER_MEMORY
        • export SPARK_EXECUTOR_INSTANCES
    • interpreter 설정
      • spark
        • master = yarn-client
        • spark.app.name = 스파크앱이름