Believe it or not, when you do a kubectl completion zsh
on your ~/.zshrc
,
kubectl
actually tries to contact the cluster of the current-context.
This is nuts and I don't understand why it would make sense.
As this can easly take seconds to complete, you can force it to quickly
give up by pointing it to a impossible HTTPS_PROXY
like this:
On your ~/.zshrc…
source <(eval HTTPS_PROXY=1:1 kubectl completion zsh)
Thank me later.