Closed consideRatio closed 1 year ago
@manics @yuvipanda @minrk what do you think? Another small major release doing this to fix our test suite and stay ahead of challenges of maintaining action-k3s-helm that setups our k8s clusters we test against?
I'm okay with that if it helps. Dropping support for k8s 1.23 seems a bit aggressive, but if it's a testing change rather than an enforced version check, that's okay with me. It's still in the GKE stable channel (and currently used in mybinder.org prod cluster).
Definitely +1 for jupyterhub version bump. Also ok with us stopping automated testing on 1.23 as long as there isn't an explicit version check for same reason @minrk mentioned.
We have released kubespawner 5.0.0 just now, but our tests testing our lower bounds of dependencies has started failing reliably for jupyterhub 1.5.1, even though they passed before, and I just can't figure out what changed.
I'd like to drop support for jupyterhub 1 and 2 for KubeSpawner, and go for
jupyterhub>=3.1.1
going onwards to not need to look into what is going wrong.I'd also like to drop testing of k8s 1.23. We have an action to setup k3s clusters that would benefit from dropping support for testing against k8s 1.23 to reduce the complexity.
Related
725