By looking at table act_id_group I don’t see a process identification, does it mean that we have to have groups that are globally unique? In other words, we can’t have Approvers group of process A and the same Approvers group of process B?
Hey Nomadus!
I might found a trick for your problem:
You could do something as written in this post right here:
So in general you can assign a task to two groups and use a Task Listener, which queries usergroups for various users using the IdentityService. So you can assign a certain task to a user who is part in both groups.
Hopefully this was helpful for your “problem”. 
Best,
Thomas
My use case is different. Let me give an example.
- Book order process, where I have a group named Approvers with one set of users;
- Book shipment process, where I have a group with the same name Approvers, but with a different set of users. How would I distinguish these groups, there is no process or tenant_id in the act_id_group table.
Ah, I understand.
I just took a look at the database schema I found here:
Taking a look on there you could possibly imagine a relation between Group and Tenant. Not sure whether this makes a lot sense but you could try and build something by yourself.
The easier possibility would be to name those groups more uniquely which I would recommend.
Naming or prefixing the group names is the default solution. I was actually hoping that there is a more elegant solution. Unfortunately, tenant_id is not in the act_id_group, so I can’t use it either and it is intended to be used in multi-org environment. If I have the same group names within an organization, it is not going to help. So naming is the only solution so far.