"$KUBEADM_TOKEN" was not of the form "\\A([a-z0-9]{6})\\.([a-z0-9]{16})\\z"

[yanxiaoqi932]

What happened?

When I set up enhanced kubernetes master:

mkdir -p /etc/kubernetes
export KUBEADM_TOKEN=`kubeadm token generate`
export APISERVER_ADDR=192.168.211.131

kubeadm init --config=/etc/kubernetes/kubeadm-client.yaml --upload-certs -v=5

it shows an error:

initconfiguration.go:306] error unmarshaling configuration schema.GroupVersionKind{Group:"kubeadm.k8s.io", Version:"v1beta3", Kind:"InitConfiguration"}: the bootstrap token "$KUBEADM_TOKEN" was not of the form "\\A([a-z0-9]{6})\\.([a-z0-9]{16})\\z"

What did you expect to happen?

run kubeadm init

How can we reproduce it (as minimally and precisely as possible)?

install enhanced k8s according to https://github.com/kubewharf/katalyst-core/blob/main/docs/install-enhanced-k8s.md

Software version

$ <software> version
# paste output here

[yanxiaoqi932]

I have got the reason. We can't use environment parameters in yaml files in this way, yaml files will think it is a string:

apiVersion: kubeadm.k8s.io/v1beta3
bootstrapTokens:
- groups:
  - system:bootstrappers:kubeadm:default-node-token
  token: $KUBEADM_TOKEN

Maybe katalyst need to fix this bug in https://github.com/kubewharf/katalyst-core/blob/main/docs/install-enhanced-k8s.md

[Pengcheng Tang]

How did you create your kubeadm-client.yaml? If you follow the documentation, it is created using heredoc:

export KUBEADM_TOKEN=`kubeadm token generate`
export APISERVER_ADDR=<your master ip address>

cat <<EOF | tee /etc/kubernetes/kubeadm-client.yaml
...
...
EOF

variable expansion should be the default behavior of heredoc

[Haiyu Zuo]

I didn't have this problem on my side, maybe there was a problem with token generate, can you use echo $KUBEADM_TOKEN to see the KUBEADM_TOKEN value?

[yanxiaoqi932]

I use vim to create kubeadm-client.yaml, this is the problem. We must use tee instead of vim because of the variable expansion of heredoc, if use vim, $KUBEADM_TOKEN will be think as a string "$KUBEADM_TOKEN" in yaml files.

[yanxiaoqi932]

Thanks, this problem has been solved. 😄