Role testing in QA environments

[Former Member]

Hi Experts,

We use the following procedure to create or modify roles:

1)Change or create the role in DEV system.

2)Generate or create the profile. At this point we create the new profile or generate the existing one, but we donu2019t use a special convention for profile names, we just accept SAP naming proposal for the profile.

3)Include the role in a transport request

4)Copy the transport request to others clients in the DEV system

5)Transport the request to QA -> PRD system.

So we have two scenarios here:

Scenario A) We often need a key user to test the role in order to make sure that all authorizations are correct and also to complain a quality process.

Scenario B) Sometimes users require to test a group of transactions in order to decide which ones should be included in a new role or in an existing role. In this point the transactions are new to the user and the transaction documentation is not enough to decideu2026. We should let them test itu2026

Now, the problem is that the DEV system (customizing client) is not a client to test, so it has no data to test.

In scenario A if we create a transport request, transport it to QA and the key user reports an authorization error, we have to create a new transport request in DEV with the corresponding fix and transport it to QA as u201Cversion 2u201D for the role.

In scenario B itu2019s possible to create a test role in DEV and proceed as the same, but when the test is finished, itu2019s necessary to create a new transport request to delete the role. This is pretty ridiculousu2026

Then, my question is:

If the test is going to take place in QA system, is it advisable to create roles in such environment for this purpose and delete them afterwards?? In this case, I should use a special naming convention for profiles to avoid problems (i.e Z_TEST<nr>). Does the role deletion in this scenario works properly? I want to make sure that problems wonu2019t arise with this role and profile creation-deletion.

And I also want to know what procedure you follow in scenarios A and B.

Hope kindly advices.

Cheers,

Diego.

[former_member204634]

Hi

In my opinion - temp role creation in quality system casts away a lot of undue process and procedures and attacks the problem straightaway - I do not if this is a best practice or not - but temp role testing works for me - before making any role changes - create a temp role in quality and get it tested and then start working on actual role in dev (one another best practice is to compare dev role with production and see if both are in sync). Hope this helps !

Prashant

[Former Member]

The trick is to be disciplined in documenting what was added when and where in the QA testing and then perform parrallel maintenance in DEV.

You must also check that the roles created in QA do not suffer from profile name collisions with DEV.

If those two are okay for you, then as Prasanth already mention, it works like a charm.

Cheers, Julius

[Former Member]

Thank You for your answers!

@Julius: I don't need parallel maintenance. I just want to create a test role in QA, generate the profile using a special naming convention. Once the role was tested the role will be deleted from QA. Then with the information gathered from the test, I want to change or create the corresponding REAL role in DEV and use the transport system to move the changes to QA and PRD.

My doubt is regarding this creation-deletion of roles. It should work properly, but I just want to know if the "cleanup" is realy complete after the deletion or if someone had problems using this strategy.

And just for a sort of survey.... How do you test the roles??

Cheers,

Diego.

[Former Member]

Then with the information gathered from the test, I want to change or create the corresponding REAL role in DEV and use the transport system to move the changes to QA and PRD.

That is what I meant by "parrallel maintenance" - you should not forget to build what did work into the real role and scrap the junkyard afterwards.

And just for a sort of survey.... How do you test the roles??

Ideal is to combine it with functional tests and I automate it to a large extent and generate the roles automatically as well as reorg them again after the changes are transfered to the real role. I work out the optimization of proposals as well and transfers the missing values to SU24 instead of the role if I find that it makes sense.

Cheers,

Julius