tencent cloud

Feedback

COS Ranger Permission System Solution

Last updated: 2022-09-29 15:37:48

    Background

    Hadoop Ranger is a permission solution for big data scenarios. By adopting the computing/storage separation mode, you can host data in COS. However, COS uses the CAM permission system, meaning that user roles and permission policies may be different from those of Hadoop Ranger. Therefore, a solution is introduced here to integrate COS with Ranger.

    Strengths

    • Fine-grained and Hadoop-compatible permission control allow you to centrally manage permissions for big data components and data hosted in the cloud.
    • There is no need to set keys in core-site on the plugin side; instead, keys are centrally set in COS Ranger Service to avoid key plaintext exposure.

    Solution Architecture

    In the Hadoop permission system, authentication is offered by Kerberos and authorized by Ranger. On the basis of this, the following components are provided to support the COS Ranger permission solution:

    1. COS Ranger Plugin: As a service defining plugin used on the Ranger server, it provides the COS service description on the Ranger side, including permission types and definitions of required parameters (such as bucket and region). Once it is deployed, you can set permission policies on the Ranger control panel.
    2. COS Ranger Service: It integrates the Ranger client, periodically syncs permission policies from the Ranger server, and verifies the permission locally after receiving an authentication request. In addition, it also provides DelegationToken generation and renewal APIs in Hadoop, all of which are defined through Hadoop IPC.
    3. Cos Ranger Client: It is dynamically loaded by the COSN plugin to forward permission verification requests to COS Ranger Service.

    Environment Deployment

    • Hadoop environment
    • ZooKeeper, Ranger, and Kerberos (if there are authentication requirements)
      Note:

      The components above are open-source and stable. You can install them on your own.

    Component Deployment

    Deploy components in the following sequence: COS Ranger Plugin, COS Ranger Service, COS Ranger Client, COSN.

    COS Ranger Plugin extends the service types in the Ranger Admin console. You can set the operation permissions related to COS in the Ranger console.

    Code address

    You can get the code from the ranger-plugin directory at GitHub.

    Version

    v1.0 or later.

    Deployment steps

    1. Create a COS directory in the service definition directory of Ranger (note: make sure that the directory permissions include at least x and r permissions).
      a. For an EMR environment, the path is ranger/ews/webapp/WEB-INF/classes/ranger-plugins.
      b. For a self-built Hadoop environment, you can find the components connected to the Ranger service through find hdfs in the ranger directory in order to find the location of the directory.
    2. Place cos-chdfs-ranger-plugin-xxx.jar (with at least the r permission) and cos-ranger.json files in the COS directory. You can get them from GitHub.
    3. Restart Ranger.
    4. Register the COS service on Ranger. You can refer to the following command:
      ## Create the service. The Ranger admin account and password as well as the Ranger service address should be specified.
      ## For an EMR cluster, the admin user is root, and the password is the root password set when the EMR cluster is created. Replace the IP of the Ranger service with the master node IP of EMR.
      adminUser=root
      ## The password set during EMR cluster creation, which is also the login password of the Ranger web service.
      adminPasswd=xxxxxx
      ## If the Ranger service has multiple master nodes, select any of them.
      rangerServerAddr=10.0.0.1:6080
      ## Specify the .json file in step 2 as -d in the command.
      curl -v -u${adminUser}:${adminPasswd} -X POST -H "Accept:application/json" -H "Content-Type:application/json" -d @./cos-ranger.json http://${rangerServerAddr}/service/plugins/definitions
      ## To delete the service just defined, you need to pass in the service ID returned during creation.
      serviceId=102
      curl -v -u${adminUser}:${adminPasswd} -X DELETE -H "Accept:application/json" -H "Content-Type:application/json" http://${rangerServerAddr}/service/plugins/definitions/${serviceId}
    5. After the service is successfully created, you can see the COS service in the Ranger console.
    6. Click + next to the COS service to define a new service instance. The service instance name is customizable; for example, you can enter cos or cos_test. The service configuration is as follows:

      You need to set the username subsequently used to start COS Ranger Service (i.e., the user allowed to pull permission policies) as policy.grantrevoke.auth.users. We generally recommend you set it to hadoop.
    7. Click the newly created COS service instance.

      Add a policy.
    8. On the page that is displayed, configure the following parameters:
    • Bucket: Bucket name, such as examplebucket-1250000000, which can be viewed in the COS console.
    • Path: Path of the COS object. Note that it does not start with a slash (/).
      • include: Indicates whether the set permission applies to the specified path itself or other paths except it.
      • recursive: Indicates that the permission applies to not only the specified path but also the subpaths under it (i.e., recursive subpaths). It is usually used when the path is set as a directory.
    • User/Group: Username and user group in logical OR relationship; that is, the operation is authorized as long as the username or user group condition is met.
    • Permissions:
      • Read: Read operation, which corresponds to the GET and HEAD operations in COS, such as downloading objects and querying object metadata.
      • Write: Write operation, which corresponds to the PUT operation in COS, such as uploading objects.
      • Delete: Deletion operation, which corresponds to the object deletion operation in COS. To rename a path in Hadoop, you need to have the deletion permission for the original path and write permission for the new path.
      • List: Traversal permission, which corresponds to the List Object operation in COS.

    Verification

    1. Use Hadoop commands to perform operations involving COSN access to check whether the current operations comply with the permissions set by the root account. Below is an example:

      # Replace the bucket, path, and other information with that of the root account.
      hadoop fs -ls cosn://examplebucket-1250000000/doc
      hadoop fs -put ./xxx.txt cosn://examplebucket-1250000000/doc/
      hadoop fs -get cosn://examplebucket-1250000000/doc/exampleobject.txt
      hadoop fs -rm cosn://examplebucket-1250000000/doc/exampleobject.txt
      
    2. Use MR Job for verification. Before the verification, be sure to restart related services such as YARN and Hive.

    FAQs

    Do I have to install Kerberos?

    Kerberos meets the authentication needs. If users in a cluster are trusted, and the purpose of the authentication is only to avoid maloperations performed by unauthorized users, you can skip installing Kerberos and only use Ranger for authentication. As a matter of fact, Kerberos also compromises the performance. Therefore, you can balance your needs for security and performance as needed. If authentication is needed, you can enable Kerberos and then configure COS Ranger Service and COS Ranger Client.

    What would happen if I enable Ranger but don't set any policy or no policy is matched?

    If no policy is matched, the operation will be denied by default.

    Can a sub-account configure the key in COS Ranger Service?

    Yes. A sub-account with relevant permissions of the manipulated bucket can generate a temporary key for the COSN plugin to perform corresponding operations. Normally, you can grant all permissions of the bucket to the configured key.

    How do I update a temporary key? Do I need to get it from COS Ranger Service every time before I access COS?

    The temporary key is cached in the COSN plugin. It will be periodically updated asynchronously.

    What should I do if the policy modified on the Ranger page doesn't take effect?

    Decrease the ranger.plugin.cos.policy.pollIntervalMs value (in milliseconds) in the ranger-cos-security.xml file and restart COS Ranger Service. After the policy is tested, we recommend you change it back to the original value (if the time interval is too short, the polling frequency will be high, causing a high CPU utilization).

    Contact Us

    Contact our sales team or business advisors to help your business.

    Technical Support

    Open a ticket if you're looking for further assistance. Our Ticket is 7x24 avaliable.

    7x24 Phone Support