WIP: hack: only repoint ipv6 kubeconfig when [::1] healthz fails - #1118
WIP: hack: only repoint ipv6 kubeconfig when [::1] healthz fails#1118杨成锴 (asjdf) wants to merge 1 commit into
Conversation
Unconditional localhost rewrite breaks kubectl on Debian-family hosts where localhost is v4-only. Ask the apiserver instead of guessing. Fixes agent-substrate#1099.
|
Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA). View this failed invocation of the CLA check for more information. For the most up to date status, view the checks section at the bottom of the pull request. |
|
|
||
| # For ipv6 kind writes a kubeconfig pointing at [::1], the address it published | ||
| # the apiserver on, which only works for a client on the Docker host itself: a | ||
| # VM-hosted daemon (Lima on macOS) forwards the port to the *v4* loopback, so |
There was a problem hiding this comment.
This seems like a bug we should report to kind upstream. Does lima expose this information in any way?
EDIT: I might argue this is a limactl bug. Seems like it should forward consistently to the macOS host.
There was a problem hiding this comment.
There was a problem hiding this comment.
cc Antonio Ojea (@aojea) we're running into kubernetes-sigs/cloud-provider-kind#422 (comment) again
There was a problem hiding this comment.
I think we might need to follow-up with limactl, I'm not sure how kind can reasonably detect this, limactl should really forward the port on the host with the same IP family?
There was a problem hiding this comment.
If we inspect the network / containers is there any indication of this there? E.g. docker desktop sets a label IIRC.
There was a problem hiding this comment.
I remember I commented somewhere about how should it work, I can not find it now
3 participants
IP_FAMILY=ipv6 hack/create-kind-cluster.shhere (no Docker);bash -nclean