-
Notifications
You must be signed in to change notification settings - Fork 19
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add information on supported k8s versions and how to pick them #19
Conversation
factored out of #923 Signed-off-by: clux <[email protected]>
Signed-off-by: clux <[email protected]>
As part of Kubernetes version policies proposed in kube-rs/website#19 for #684 Signed-off-by: clux <[email protected]>
Signed-off-by: clux <[email protected]>
Have factored this out of the general policy PR and cleaned it up a bit. It is kind of mergeable together with kube-rs/kube#924. But could use some thoughts on the points edited into the main PR. |
Signed-off-by: clux <[email protected]>
Signed-off-by: clux <[email protected]>
Signed-off-by: clux <[email protected]>
OK, have added a small bit about how we can remove special abstractions when the upstream source removes them to avoid being forced into a double support period for things like CRD/v1beta1. All in all, I am reasonably happy with this now. The policy, while a bit arbitrary is easy to implement, matches real world use cases (cloud providers), and it explains what to expect + our special cases. Will merge later tomorrow unless there are no comments on this. |
Co-authored-by: kazk <[email protected]> Signed-off-by: Eirik A <[email protected]>
Co-authored-by: kazk <[email protected]> Signed-off-by: Eirik A <[email protected]>
Co-authored-by: kazk <[email protected]> Signed-off-by: Eirik A <[email protected]>
Co-authored-by: kazk <[email protected]> Signed-off-by: Eirik A <[email protected]>
Co-authored-by: kazk <[email protected]> Signed-off-by: Eirik A <[email protected]>
man, i am really bad with those capitalizations huh 😅 Thanks! |
* Run e2e tests against supported kubernetes versions As part of Kubernetes version policies proposed in kube-rs/website#19 for #684 Signed-off-by: clux <[email protected]> * add integration tests as specific job to run it against mk8sv Signed-off-by: clux <[email protected]> * rename tests and try to fix integration Signed-off-by: clux <[email protected]> * fix k3d args? Signed-off-by: clux <[email protected]> * fix e2e Signed-off-by: clux <[email protected]> * test against v1.19 Signed-off-by: clux <[email protected]> * badge update Signed-off-by: clux <[email protected]> * s/MINK8SV/MK8SV Signed-off-by: clux <[email protected]> * Big e2e rework (WIP) Now two e2e tests, and more feature combinations. Signed-off-by: clux <[email protected]> * avoid rustls local due to the issues Signed-off-by: clux <[email protected]> * split e2e tests into own file and clean up when to matrix Signed-off-by: clux <[email protected]> * running bump-k8s also bumps mk8sv Signed-off-by: clux <[email protected]> * avoid illegal anchors and add badge bumping Signed-off-by: clux <[email protected]> * better version checks for mk8sv and msrv on ci Signed-off-by: clux <[email protected]> * move some ci around for consistency in naming Signed-off-by: clux <[email protected]> * stray command Signed-off-by: clux <[email protected]> * fix syntax errors? Signed-off-by: clux <[email protected]> * fix test warning from kube-derive after apiext removal Signed-off-by: clux <[email protected]> * evars from tests Signed-off-by: clux <[email protected]> * better run description on integration jobs Signed-off-by: clux <[email protected]> * better output for flaky integration test Signed-off-by: clux <[email protected]> * fmt + give test more time? Signed-off-by: clux <[email protected]> * sanity after timeout. seems likely cause. Signed-off-by: clux <[email protected]> * no need for an agent in k3d cmd Signed-off-by: clux <[email protected]> * how slow is this actions cluster? Signed-off-by: clux <[email protected]> * re-trigger once more to check * apiserver, are you ok? Signed-off-by: clux <[email protected]> * another empty commit to retrigger Signed-off-by: clux <[email protected]> * remove duplicate changes from #924 (won't build until that is merged) Signed-off-by: clux <[email protected]>
Signed-off-by: clux <[email protected]> link to future stability policy discussion Signed-off-by: clux <[email protected]> create a kubernetes versioning matrix now since we can't use kubernetes versioning post 1.0, there's no point waiting. let's define and get clarity now. Signed-off-by: clux <[email protected]> potential solution for kubernetes version policy Propose a 5 version range from [MAX-4, MAX], since this matches kubernetes upstream support. This will need to go with better e2e tests. Signed-off-by: clux <[email protected]> remove kubernetes doc - moved into #19 Signed-off-by: clux <[email protected]> update with unstable features Signed-off-by: clux <[email protected]>
Kubernetes versioning docs (factored out of #18):
Done
Slightly iffy
v1_19
, v1_24], but it was actually removed from structs in
v1_22`)For the badge; we link to this policy document because it's unclear what a soft minimum means. MK8SV is always going to be a soft minimum, and what matters is how that range changes what you can expect. As a result, have also mentioned the stricter part about when we remove our own special abstractions (like CRD/v1beta1).
Unfortunately we cannot always test the maximum directly (k3s releases lagging as found in kube-rs/kube#924 ) but at least that PR does a good job of improving the situation.