Layer7 API Management

Expand all | Collapse all

Inconsistent Policy Code in Policy Manager

  • 1.  Inconsistent Policy Code in Policy Manager

    Posted 01-06-2017 12:27 PM

    In our team we are facing big issues with inconsistency of policy code. 

     

    For example:

    I am user A:Login to Policy Manager creates a Policy 'abc' with some assertions.

    My teammate user B: Logsin to Policy Manager with his credentials. He opens same policy 'abc' created by User A.

    Perform some changes example add's new compare expression or more assertions to build a logic in the policy to fix some defects. 

    Saves and Activates the code and unit tested using SoapUI and policy is behaving as expected. 

     

    Next User A logsin -> Refreshes his Policy Manager. 

    User A opens the same policy 'abc' and he couldn't see the changes made by User B.

    User B also login at the same time but he can see the changes. 

     

    If User A perfroms some change without validating the changes made by User B and activiates the policy then the changes made by User B is lost. 

     

     

    I am not sure why this inconsistent behavior. Are we need to perform some steps to have a uniform visibility of code?

     

    As per my understanding all policies are resides in Gateway and policy manager retrieves the files and publishes to the editor.

     

    There should be one view once code is Saved and published to the gateway. Can someone please help to understand the this?

     

     

     

    Regards

    Kareem



  • 2.  Re: Inconsistent Policy Code in Policy Manager

    Posted 01-06-2017 06:53 PM

    Kareem,

     

    The information that the client would be pulling to the Policy Manager is from the database directly when the policy is opened. The visibility of the assertions or versions of the policy are not controlled by an account by account basis. The scenario that you have outlined is not expected behavior. Are the services and policies in a cluster or a stand alone node? If they are in a cluster does the client connect to a VIP or an individual node in the cluster? The reason I ask we have seen when a cluster is incorrectly configured with each node pointing to it's own db as the primary then replication will fail it will cause a inconsistency between the nodes in the cluster. Always ensure that both nodes in the cluster are pointing to the same primary DB node as their primary and the secondary DB as their secondary.

     

    Also what version of the product are you using?

     

    Sincerely,

     

    Stephen Hughes

    Director, CA Support