Hi spicedb team! Any chance ya'll publish Kuberentes version compatibility somewhere for the

spicedb-operator

? I can sorta infer it from the client-go import, but figured I'd ask in case it's documented somewhere I can reference. Thanks!

Cc @ecordell

and when you see the above, i'm also curious if the operator supports

> 1

replica. I bumped the

Deployment

replicas and the pods run, but from some quick code searching I think there is no leaderelection in place, and just wanted to verify 🙂 . Have a good weekend!

I don't think it's designed to work with more than 1 replica, there is no leader election happening. If the operator goes down, you spiceDB cluster will continue to operate normally.

Yup! Makes sense -- thanks for verifying. I'll keep an eye out if anyone knows the answer about the kube-version.

Yeah, no leader election at the moment, it doesn’t really need to be HA (but if you have a concern and want to run it HA feel free to open an issue and we can look at adding leader election). The operator tests run again a kube 1.25.3, and we have it running in test/stage/prod environments up to 1.29 (I think 1.30 too but would need to check). It probably works on much older versions just fine If you’re looking for something more formal we can work on that; my expectation is we’ll have some minimum version and then test new versions as they come out.

> Yeah, no leader election at the moment, it doesn’t really need to be HA (but if you have a concern and want to run it HA feel free to open an issue and we can look at adding leader election). Generally I think you're right and it's not high prio. If anything just checking for my own understanding. Probably the only risk is that you can scale up the replicas as is (as I tested) -- I wonder if the operators when then race to try and manage clusters per CRDs. Again, not a huge issue, mostly curious. > If you’re looking for something more formal we can work on that; my expectation is we’ll have some minimum version and then test new versions as they come out. The most formal thing I'd like is it laid out in docs somewhere. It just makes it easier to reference when we're doing upgrades and such. But this all sounds good! Appreciate the follow-up. Unrelated, I think we both worked at CoreOS at the same time -- good to see you name in here!