Join GitHub today
GitHub is home to over 28 million developers working together to host and review code, manage projects, and build software together.Sign up
kubeadm: add configuration option to not taint master #55479
What this PR does / why we need it:
Although tainting the master is normally a good and proper thing to do in some situations (docker for mac in our case, but I suppose minikube and such as well) having a single host configuration is desirable.
In linuxkit we have a workaround to remove the taint after initialisation. With the change here we could simply populate
I have only added this to the config file and not to the CLI since AIUI the latter is somewhat deprecated.
The code also arranges to remove an existing taint if it is unwanted. I'm unsure if this behaviour is correct or desirable, I think a reasonable argument could be made for leaving an existing taint in place too.
Signed-off-by: Ian Campbell email@example.com
Since the requirement for this option is rather niche and not best practice in the majority of cases I'm not sure if it warrants mentioning in the release notes? If it were then perhaps
I'm sorry @ijc, it might be too late to get this into v1.9 at this time I think.
If you can make the SIG meeting tomorrow or the kubeadm meeting on Weds though; we might be able to briefly discuss it and prioritize accordingly.
I'm sorry that no-one responded to this earlier, but we didn't get notified. ^ is the Github team tag to use for getting the notification across to the right people
Sorry, I never made it to the Wednesday meeting last year, things were a bit hectic.
Is now the right time in the v1.10 release cycle to revisit this? (I think/hope so based on kubeadm).
Despite being 1500+ commits behind this still appears to still merge cleanly. Perhaps I should rebase nonetheless? (I'll check it still builds when merged with master in any case)
cc @luxas @kubernetes/sig-cluster-lifecycle-pr-reviews
In response to this:
Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.
I've manually run though using config file with
I've not run through the manual
Tainting was as expected/required at each step.
@timothysc I figured minikube and docker for mac between them were not that all that exceptional and a post provision step is one more thing to do wrong/differently or to bit rot. d4m would certainly use this functionality if it were available (I don't know about minikube, I'm only speculating that it would be useful for that use case too).
I'll hold off on the rebase until we've decided whether or not this is a worthwhile change.
[APPROVALNOTIFIER] This PR is APPROVED
The full list of commands accepted by this bot can be found here.
The pull request process is described here
Needs approval from an approver in each of these OWNERS Files:
Approvers can indicate their approval by writing