Using Hadoop from Scratch

Procedure

1. Create a cluster.
   1. Log in to the management console.
   2. Choose EI Enterprise Intelligence > MapReduce Service . The MRS management console is displayed.
   3. Click Buy Cluster, the Buy Cluster page is displayed.
   4. Click the Custom Config tab.

2. Configure software.
   1. Region : Select a region.
   2. Billing Mode : Select Pay-per-use .
   3. Cluster Name : Enter mrs_demo or use another name following the naming rules.
   4. Cluster Type : Select Analysis cluster .
   5. Cluster Version : Select MRS 3.1.0 .
   6. Component : Select all components required for an analysis cluster.
   7. Click Next .

3. Configure hardware.
   1. AZ : Select AZ2 .
   2. Enterprise Project : Select default .
   3. VPC and Subnet : Retain their default values or click View VPC and View Subnet to create new ones.
   4. Subnet : Retain the default value.
   5. Security Group : Use the default option Auto create .
   6. EIP : Retain the default option Bind later .
   7. Cluster Nodes : Retain the default settings. Do not add new Task nodes.
   8. Click Next .

4. Set advanced options.
   1. Kerberos Authentication : Trun off this function.
   2. Username : admin is used by default.
   3. Password and Confirm Password : Set them to the password of the FusionInsight Manager administrator.
   4. Login Mode : Select Password . Enter a password and confirm the password for user root .
   5. Hostname Prefix : Retain the default value.
   6. Select Configure for Set Advanced Options , and set Agency to MRS_ECS_DEFAULT_AGENCY .
   7. Click Next .

5. Confirm the configuration.
   1. Configure : Confirm the settings configured on the Configure Software , Configure Hardware , and Set Advanced Options pages.
   2. Enable Secure Communications .
   3. Click Buy Now. A page is displayed showing that the task has been submitted.
   4. Click Back to Cluster List . You can view the status of the newly created cluster on the Active Clusters page. Wait for the cluster creation to complete. The initial status of the cluster is Starting . After the cluster is created, the cluster status becomes Running .

6. Prepare the Hadoop sample program and data files.
   1. Prepare the wordcount program.
      1. Log in to the Master1 node as user root .
      2. Run the following commands to go to the client installation directory, configure environment variables, and create an HDFS directory, for example, /user/example :

         cd /opt/Bigdata/client

         source bigdata_env

         hdfs dfs -mkdir /user/example

      3. Run the following commands to upload the /opt/Bigdata/client/HDFS/hadoop/share/hadoop/mapreduce/hadoop-mapreduce-examples-3.1.1-*.jar file to the /user/example directory: Replace the * with the JAR version you are using.

         cd HDFS/hadoop/share/hadoop/mapreduce

         hdfs dfs -put hadoop-mapreduce-examples-3.1.1-*.jar /user/example

   2. Prepare data files.

      There is no requirement on the format of the data files. Prepare two .txt files. For example, files wordcount1.txt and wordcount2.txt .

      Figure 1 Preparing data files

      
      

7. Upload files to OBS.
   1. Log in to the OBS console and choose Parallel File Systems . On the Parallel File Systems page, click Create Parallel File System . On the displayed page, create a file system named mrs-word01 .
   2. Click the name of the mrs-word01 file system. In the navigation pane on the left, choose Files. On the page that is displayed, click Create Folder to create two folders: program and input .
   3. Go to the input folder and upload the wordcount1.txt and wordcount2.txt files prepared in 6.
   4. In the navigation pane of the MRS console, choose Active Clusters. On the Active Clusters page, click the mrs_demo cluster.
   5. In the cluster list, click a cluster name and click the Files tab on the cluster details page. On the displayed page, click Export Data . In the Export Data from HDFS to OBS dialog box, set the following parameters and click OK :
      • HDFS Path : Select the path of the prepared JAR package, that is, /user/example/hadoop-mapreduce-examples-3.1.1-*.jar / .
      • OBS Path : Select the path of the program folder, that is, obs://mrs-word01/program . Select the check box for confirming script security and click OK .
   6. To submit a job on the GUI, go to 8. To submit a job through a cluster node, go to 9.

8. Submit a job on the GUI.
   1. On the cluster information page, click the Jobs tab then Create to create a job.
   2. In the Create Job dialog box, set the following parameters:
      • Type : Choose MapReduce .
      • Job Name : Enter wordcount .
      • Program Path : Click OBS and select the Hadoop sample program uploaded in 7.
      • Parameters : Enter wordcount obs://mrs-word01/input/ obs://mrs-word01/output/ .

        obs://mrs-word01/output/ indicates the output path, which is a directory that does not exist.

      • Service Parameter : Leave it blank.
   3. Click OK to submit the job. After a job is submitted, it is in the Accepted state by default. You do not need to manually execute the job.
   4. Go to the Jobs tab to view the job status and logs. Then go to 10 to view the job execution result.

9. Submit a job through a cluster node.
   1. Log in to the MRS console and click the cluster named mrs_demo to go to its details page.
   2. Click the Nodes tab. In this tab, click the name of the master1 node to go to the ECS management console.
   3. Click Remote Login in the upper right corner of the page.
   4. Enter the username and password of the master node as prompted. The username is root and the password is the root password configured during cluster creation.
   5. Run the following command to configure environment variables:

      cd /opt/Bigdata/client

      source bigdata_env

   6. Run the following command to submit the wordcount job, read data from OBS, and output the execution result to OBS:

      hadoop jar HDFS/hadoop/share/hadoop/mapreduce/hadoop-mapreduce-examples-3.1.1-*.jar wordcount "obs://mrs-word01/input/*" "obs://mrs-word01/output/"

      obs://mrs-word01/output/ indicates a path for storing job output files on OBS and needs to be set to a directory that does not exist.

10. Query job execution results.
    1. Log in to OBS console and click the name of the mrs-word01 parallel file system.
    2. On the page that is displayed, choose Files in the navigation pane on the left. Go to the output path in the mrs-word01 bucket specified during job submission, and view the job output file. You need to download the file to a local computer and open it in .txt format.