Apache Ambari Views
Also available as:
PDF
loading table of contents...

Configuring Your Cluster for Workflow Manager View

For Workflow Manager View to access HDFS, the Ambari Server daemon hosting the view must act as the proxy user for HDFS. This allows Ambari to submit requests to HDFS on behalf of the Worklow Manager View users.

Each Worklow Manager View user must have a user directory set up in HDFS.

If the cluster is configured for Kerberos, ensure that the the section called “Configuring Views for Kerberos” has been completed.

Set up HDFS Proxy User
[Note]Note

If you previously set up the proxy user for another View, you can skip this task.

To set up an HDFS proxy user for the Ambari Server daemon account, you need to configure the proxy user in the HDFS configuration. This configuration is determined by the account name the ambari-server daemon is running as. For example, if your ambari-server is running as root, you set up an HDFS proxy user for root.

Steps

  1. In Ambari Web, browse to Services > HDFS > Configs.

  2. Under the Advanced tab, navigate to the Custom core-site section.

  3. Click Add Property… to add the following custom properties:

    hadoop.proxyuser.root.groups="users"
    hadoop.proxyuser.root.hosts=ambari-server.hostname

    Notice the ambari-server daemon account name root is part of the property name. Be sure to modify this property name for the account name you are running the ambari-server as. For example, if you were running ambari-server daemon under an account name of ambariusr, you would use the following properties instead:

    hadoop.proxyuser.ambariusr.groups="users"
    hadoop.proxyuser.ambariusr.hosts=ambari-server.hostname

    Similarly, if you have configured Ambari Server for Kerberos, be sure to modify this property name for the primary Kerberos principal user. For example, if ambari-server is setup for Kerberos using principal ambari-server@EXAMPLE.COM, you would use the following properties instead:

    hadoop.proxyuser.ambari-server.groups="users"
    hadoop.proxyuser.ambari-server.hosts=ambari-server.hostname
  4. Save the configuration change and restart the required components as indicated by Ambari.

More Information

Ambari Server for Kerberos

Set up HDFS User Directory

You must set up a directory for each user that accesses the Workflow Manager View. Workflow Manager View stores user metadata in the user directory in HDFS.

About This Task

By default, the location in HDFS for the user directory is /user/${username}, where ${username} is the username of the currently logged in user that is accessing Workflow Manager View.

[Important]Important

Since many users leverage the default Ambari admin user for getting started with Ambari, you should create a /user/admin directory if one does not exist, in addition to directories for other Workflow Manager View users.

Steps

  1. Connect to a host in the cluster that includes the HDFS client.

  2. Switch to the HDFS system account user.

    su - hdfs

  3. If working on a secure Kerberos cluster:

    1. Destroy any existing Kerberos tickets:

      kdestroy

      If no ticket is found, you get an error message that you can ignore: No credentials cache file found while destroying cache

    2. Obtain a Kerberos ticket-granting ticket:

      kinit -kt /etc/security/keytabs/hdfs.headless.keytab hdfs

  4. Using the HDFS client, make an HDFS directory for the user.

    For example, if your username is wfmadmin, you would create the following directory.

    hadoop fs -mkdir /user/wfmadmin

  5. Set the ownership on the newly created directory.

    For example, if your username is wfmadmin, the directory owner should be wfmadmin:hadoop.

    hadoop fs -chown wfmadmin:hadoop /user/wfmadmin

  6. Log in as root user.

    su -

  7. Access the Kerberos administration system.

    kadmin.local

  8. Create a new principal and password for the user.

    You can use the same user name and password that you used for HDFS directory.

    addprinc -pw [wfmadmin-password] wfmadmin@EXAMPLE.COM

  9. Repeat steps 2 through 8 for any additional Workflow Manager View users.