Is there an existing issue for this?
Current Behavior
Take an valid empty kubeconfig, then try to render a bundle locally. Fleet CLI is supposed to work with no k8s cluster accessible:
% cat kubeconfig-empty.yaml
apiVersion: v1
kind: Config
preferences: {}
clusters: []
users: []
contexts: []
% env KUBECONFIG=kubeconfig-empty.yaml ./fleet-linux-amd64-0.15.4 apply -o bundle.yaml some ./some || echo "rc=$?"
rc=1
% env KUBECONFIG=kubeconfig-empty.yaml ./fleet-linux-amd64-0.16.0-rc.4 apply -o bundle.yaml some ./some || echo "rc=$?"
rc=1
% ls -l bundle.yaml
"bundle.yaml": No such file or directory (os error 2)
% ./fleet-linux-amd64-0.15.4 --version; ./fleet-linux-amd64-0.16.0-rc.4 --version
fleet version v0.15.4 (7b97db764df7de385e0572eab6da959e0bd22569)
fleet version v0.16.0-rc.4 (4b6976899c8cfc16317e7e5d610c74caa56ddd34)
It is working fine with a dummy cluster added:
% cat kubeconfig-dummy.yaml
apiVersion: v1
kind: Config
preferences: {}
clusters:
- {name: dummy, cluster: {server: whatever}}
contexts:
- {name: dummy, context: {cluster: dummy}}
current-context: dummy
% env KUBECONFIG=kubeconfig-dummy.yaml ./fleet-linux-amd64-0.15.4 apply -o bundle.yaml some ./some || echo "rc=$?"
% ls -l bundle.yaml
.rw------- 86k user 10 Jul 13:30 bundle.yaml
Expected Behavior
fleet apply -o bundle.yaml … should produce output with valid empty kubeconfig.
Steps To Reproduce
All info is in the "Current behaviour" section.
Environment
- Architecture: amd64
- Fleet Version: 0.15.4, 0.16.0-rc.4
- Cluster: N/A
Logs
Anything else?
It seems the CLI tries to detect whether it is run locally or in a pod and fails:
% env KUBECONFIG=kubeconfig-empty.yaml strace -et=%file -es=none -eq=all -s99999 -f ./fleet-linux-amd64-0.15.4 apply -o bundle.yaml some ./some
…
[pid 1617654] openat(AT_FDCWD, "kubeconfig-empty.yaml", O_RDONLY|O_CLOEXEC) = 3
[pid 1617654] newfstatat(AT_FDCWD, "/var/run/secrets/kubernetes.io/serviceaccount/token", 0xc0007cd628, 0) = -1 ENOENT (No such file or directory)
It shouldn't do that check if run with -o / --output option.
Is there an existing issue for this?
Current Behavior
Take an valid empty kubeconfig, then try to render a bundle locally. Fleet CLI is supposed to work with no k8s cluster accessible:
It is working fine with a dummy cluster added:
Expected Behavior
fleet apply -o bundle.yaml …should produce output with valid empty kubeconfig.Steps To Reproduce
All info is in the "Current behaviour" section.
Environment
Logs
Anything else?
It seems the CLI tries to detect whether it is run locally or in a pod and fails:
It shouldn't do that check if run with
-o/--outputoption.