This message was deleted.

Hello @Adrian Patterson , I’ll check with the team members and get back to you

Hi @Adrian Patterson :)

An OPAL server can pull policies from more than one git repo

• You can do that via OPAL scopes - see here: https://docs.opal.ac/overview/scopes

An OPAL client can connect to more than one OPAL server and specifiy the policyDirs for each server

A client can move between servers- but there’s no reason to do that. Instead you can have the client subscribe to multiple scopes /topics (topics translate to different directories, each of those topics would propagate between the servers - the servers should be connected via a backbone pus/sub(e.g. Redis, Kafka, Postgres)

Thank you Or! I will look into scopes, it sounds like what we would need. As always thank you guys for your quick support :)

At the moment I believe it’s one scope per client, but you can have multiple topics (directories) within that scope. (@Asaf Cohen , @Ori Shavit - correct me please if I’m wrong) If you think we should add multiple scopes per client; please open an issue for it in Github

I'm just thinking back to my scenario, if I want policies from two different repos as an OPAL client but I can only use one scope, then I can only get bundles from the single repo?

In short, yes- unless we add multi-scopes per client (BTW you can also contribute that on your own to the project- we’ll help you) To offer an alternative: You can consider merging the repos upstream of OPAL into an intermediate repo (and manage it with something like Git submodules for example) You can even have one main merged repo for all the clients (So even just one scope - or even without using scopes) -and as you merge things into separate folders in that repo- you can use topics (aka policy dirs) to have different combo for different clients.

Gotcha. Do you think this would be a feature people would be interested in? As context, our optimal scenario is where we have a central generic policy repo which applies company wide in different projects, then sub projects with their own policies. Clients can then use policies from both repos.

Well this is the first time this has come up; so it’s hard to say; I think it makes general sense - and probably be useful to more users than just you; though I don’t think it would be a very common scenario.

Fair enough, I couldn't find anything for my specific scenario when researching. Alright then! I'll look into alternatives/how complex it would be to add multiple scopes per client. Thanks again for all your help Or, hope you have a great day 😊

Please do share your findings 🙂 Wishing you a great day as well :)