| | 1 | [[PageOutline]] |
| | 2 | |
| | 3 | ◢ <[wiki:BigThingCamp160227/Lab1 實作一]> | <[wiki:BigThingCamp160227 回課程大綱]> ▲ | <[wiki:BigThingCamp160227/Lab2 實作二]> ◣ |
| | 4 | |
| | 5 | = 實作一 Lab 1 = |
| | 6 | |
| | 7 | {{{ |
| | 8 | #!html |
| | 9 | <div style="text-align: center;"><big style="font-weight: bold;"><big>在 Koding.com 上安裝 Hadoop<br/>Hadoop installation in Practice</big></big></div> |
| | 10 | }}} |
| | 11 | |
| | 12 | == STEP 0 : 註冊 koding 帳號並開啟一台虛擬機器 == |
| | 13 | |
| | 14 | * 連線至 https://koding.com/Home 選擇 Sign Up 註冊新的帳號,或者用 github 帳號登入 |
| | 15 | * 連上虛擬機器 https://koding.com/Terminal |
| | 16 | |
| | 17 | == STEP 1 : 從 github 取得本次課程的範例 == |
| | 18 | |
| | 19 | * 在 Terminal 中輸入以下指令 |
| | 20 | {{{ |
| | 21 | ~$ git clone https://github.com/jazzwang/hadoop_labs.git |
| | 22 | }}} |
| | 23 | |
| | 24 | * 您應該會看到類似底下的結果: |
| | 25 | {{{ |
| | 26 | #!text |
| | 27 | ~$ git clone https://github.com/jazzwang/hadoop_labs.git |
| | 28 | Cloning into 'hadoop_labs'... |
| | 29 | remote: Counting objects: 249, done. |
| | 30 | remote: Compressing objects: 100% (166/166), done. |
| | 31 | remote: Total 249 (delta 115), reused 176 (delta 44) |
| | 32 | Receiving objects: 100% (249/249), 53.64 KiB, done. |
| | 33 | Resolving deltas: 100% (115/115), done. |
| | 34 | }}} |
| | 35 | * 檢查是否有 hadoop_labs 目錄 |
| | 36 | {{{ |
| | 37 | ~$ cd hadoop_labs/ |
| | 38 | ~/hadoop_labs$ ls -al |
| | 39 | }}} |
| | 40 | |
| | 41 | == STEP 2 : 執行安裝腳本 == |
| | 42 | |
| | 43 | * 首先, 我們來介紹 Hadoop 的三種安裝模式 |
| | 44 | * <參考> http://hadoop.apache.org/docs/stable/single_node_setup.html |
| | 45 | {{{ |
| | 46 | #!text |
| | 47 | Now you are ready to start your Hadoop cluster in one of the three supported modes: |
| | 48 | |
| | 49 | * Local (Standalone) Mode |
| | 50 | * Pseudo-Distributed Mode |
| | 51 | * Fully-Distributed Mode |
| | 52 | }}} |
| | 53 | |
| | 54 | * 開始動手吧~請剪貼以下的步驟: |
| | 55 | {{{ |
| | 56 | ~$ cd ~/hadoop_labs |
| | 57 | ~/hadoop_labs$ sudo apt-get -y install wget |
| | 58 | ~/hadoop_labs$ lab000/hadoop-local-mode |
| | 59 | }}} |
| | 60 | |
| | 61 | * 等待安裝的過程中,讓我們來講解 [https://raw.github.com/jazzwang/hadoop_labs/master/lab000/hadoop-local-mode hadoop-local-mode 這隻 Shell Script] 做了哪些事情。 |
| | 62 | 1. 安裝 Java Runtime Environment (JRE) 與 Java Development Kit (JDK) - 雖然目前 Oracle 已經釋出 JDK/JRE7,但 JDK/JRE 6 還是 Hadoop 開發者有經過大量測試驗證的版本。未來若要進行商業運轉,建議安裝 CDH4 或 HDP 搭配 JRE7。 |
| | 63 | 2. 下載 hadoop-$VERSION.tar.gz |
| | 64 | 3. 解壓縮到 ${HOME}/hadoop |
| | 65 | 4. 設定 ${HOME}/hadoop/conf.local/hadoop-env.sh |
| | 66 | 5. 設定 ${HOME}/.bashrc 加入 PATH 環境變數 |
| | 67 | |
| | 68 | * 安裝完成,首先先讓我們觀察有幾個 java process |
| | 69 | {{{ |
| | 70 | ~/hadoop_labs$ jps |
| | 71 | }}} |
| | 72 | |
| | 73 | * 觀察有沒有開 port |
| | 74 | {{{ |
| | 75 | ~/hadoop_labs$ netstat -nap | grep java |
| | 76 | }}} |
| | 77 | |
| | 78 | * 讓我們來複習一下 HDFS 的基本操作 |
| | 79 | {{{ |
| | 80 | ~/hadoop_labs$ cd ~ |
| | 81 | ~$ ls |
| | 82 | ~$ source ~/.bashrc |
| | 83 | ~$ hadoop fs -ls |
| | 84 | ~$ hadoop fs -mkdir tmp |
| | 85 | ~$ hadoop fs -ls |
| | 86 | ~$ ls |
| | 87 | ~$ hadoop fs -put ${HOME}/hadoop/conf.local input |
| | 88 | ~$ hadoop fs -ls |
| | 89 | ~$ ls |
| | 90 | }}} |
| | 91 | |
| | 92 | == 實作習題 == |
| | 93 | |
| | 94 | 試回答以下問題: |
| | 95 | |
| | 96 | 1. 當您下 jps 指令之後,除了 jps 以外,看到幾個 java process ? |
| | 97 | {{{ |
| | 98 | #!text |
| | 99 | (A) 0,只有 jps 一個結果 |
| | 100 | (B) 1 |
| | 101 | (C) 2 |
| | 102 | (D) 3 |
| | 103 | (E) 4 |
| | 104 | }}} |
| | 105 | 2. 回到家目錄,執行 hadoop fs -ls 指令,您看到的結果跟直接下 ls 有無不同? |
| | 106 | {{{ |
| | 107 | #!text |
| | 108 | (A) hadoop fs -ls 會顯示完整路徑,並將隱藏檔案也秀出來. ls 只會秀出一般的檔案 |
| | 109 | (B) 顯示 can not access |
| | 110 | }}} |
| | 111 | 3. 在家目錄, 執行 hadoop fs -mkdir tmp 指令,對家目錄有何影響? |
| | 112 | {{{ |
| | 113 | #!text |
| | 114 | (A) 在家目錄底下多了一個 tmp 目錄 |
| | 115 | (B) 家目錄沒有任何改變 |
| | 116 | }}} |