Joel Speed
Joel Speed
/lgtm /label backport-risk-assessed
> However there will be a downside - we can't in-place change (add / remove) taints for those worker nodes Removing taints becomes problematic if you have the potential for...
> If taints are added later Pods that don't tolerate the taint might have already been scheduled to the Node To be fair, that is the purpose of the NoExecute...
> Makes sense. I guess the only limitation then is that if you add the taints later you cannot ensure that the Pods never ran on these Nodes So this...
I think given that users of Kubernetes without Cluster API would expect taints on Nodes to be mutable without having to replace the Nodes, if a user modified the taints...
One note, `listType=set` is ok for `[]string`, but has issues if you try and use `[]struct{}` in any way, I generally recommend avoiding `listType=set` in general and prefer CEL Some...
Hmm interesting problem, so, what issue does this actually cause when there's a mismatch? And what would you expect the behaviour to be here? How would you want to be...
This feels like something that is somewhat an opinion (a strong opinion I agree with), but not necessarily required. I could call my package anything, but call my API version...
This makes sense, but I wonder, if we ever changed the XValidation priority, would we pick up on this? Is there a way to make sure in the future that...