Kubectl on MacOS Montery 12.6 causing networking crash, requires restart

[awbacker]

When I execute kubectl version on my recently upgraded OSX 12.6 Macbook Pro (Intel), the network stack appears to crash. This requires a restart.

This appears to be related to IPv6 in some way:

• Wifi DHCP provies 2 servers: 192.168.1.254 and fe80::c6e5:32ff:...
• Manually setting to 8.8.4.4 or another IPv4 fixes the issue

Perhaps kubectl is triggering some MacOS network stack bug?

What happened:

Unable to connect to the server: dial tcp: lookup {...}.eks.amazonaws.com: i/o timeout

• The entire network stack stops responding
• Network settings page blanks out completely
• kubectl version can't be exited with ctrl+c
• kill -9 on the kubectl process results in (kubectl) in ps aux output, it never leaves
• Restarting causes MacOS to force quit apps such as firefox, they can't shutdown cleanly

What you expected to happen:

Expected it would show the version.

Environment:

• Connecting to AWS EKS
• Kubernetes server v1.21
• kubectl clients v1.21, v1.24, v1.25
• Tested with AWS's provided EKS as well as official distributions
• Mac OS 12.6, Intel

I found a comment in an issue in helm (helm/helm#11306 (comment)) which solved this for me: Change to IPv4 only in DNS.

I was running into the same issue and discovered it was being caused by a weird IPv6 DNS record in my System Settings. These greyed out rules were set automatically (based on my network setup I think), but after manually setting just 192.168.1.254 as my sole DNS server my issues were resolved instantly. System Preferences -> Network -> Advanced -> DNS

[k8s-ci-robot]

@awbacker: This issue is currently awaiting triage.

SIG CLI takes a lead on issue triage for this repo, but any Kubernetes member can accept issues by applying the triage/accepted label.

The triage/accepted label can be added by org members by writing /triage accepted in a comment.

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.

[dims]

@awbacker you mentioned Related issue I ran into, which gave me the solution... can you please comment what exactly you had to do?

[awbacker]

@dims

Manually setting to 8.8.4.4 or another IPv4 fixes the issue

The specific comment is helm/helm#11306 (comment). I've updated the issue with a link to the comment directly, and a copy of the text posted by that user. I can see how my "that fixed it" would be pretty easy to miss.

[eddiezane]

This appears to be an upstream Go issue. Let's follow it there.

golang/go#52839

/close

[k8s-ci-robot]

@eddiezane: Closing this issue.

In response to this:

This appears to be an upstream Go issue. Let's follow it there.

golang/go#52839

/close

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.