User from AD and group in Internal directory

[Rilwan Ahmed]

We are planning to fetch all users from Main Active Directory and create groups in Internal directory. 
As of now, jira has groups from AD. Incase if I create groups in Internal directory similar to that of AD groups along with users and remove the existing AD groups (https://confluence.atlassian.com/jirakb/synchronize-users-only-without-any-groups-from-active-directory-into-jira-387941285.html) ,  will the roles, permission scheme, global permisisons etc using link to AD group will get affected  or will look for the same AD names from Internal directory?

[Andy Heinzer]

This can be difficult to answer.  Having two different groups with the same name that originate from different user directories can be a problem for Jira to handle.  

When you grant access to a specific group in Jira, there are database tables that get updated to reflect this.  But my understanding is that these groups are being generated unique identifiers that pertain to the user directory they come from.

So if you were to create the same group in both the AD directory and the internal directory, I typically would not expect the permissions to aggregate groups across directories.  This can be confusing, because I think that Confluence/Crowd might actually be able to do that.   But in terms of Jira, I find it tends to be better to create distinctly unique group names per directory as a means to clarify where this directory exists.  Jira does not do a great job of explaining which user directory a group come from when you are adding a group to a permission/role/etc.  So there are possible permissions losses when you remove the old groups.

But honestly, I could be wrong about this because things change, applications update, and maybe the latest versions handle this better now.   I would recommend creating a staging instance of your Jira.  This would allow you to test this out without doing so on your production instance first.  It might seem like more work up front to test like this, but potentially I think this could save you a lot of time in order to determine exactly how Jira will respond to the changes I think you are looking to make here.