Fix libvirt-coreos cluster #11379
Fix libvirt-coreos cluster #11379
Conversation
KUBERNETES=libvirt-coreos cluster/kube-up.sh produced the following error: cluster/../cluster/libvirt-coreos/../../cluster/common.sh: line 83: user_args[@]: unbound variable This was coming from the fact that, as a libvirt-coreos cluster runs locally on local VMs, there is no authentication mechanism. This led to have user_args of common.sh unset. In the case of libvirt-coreos, it is in fact expected to have no authentication token.
|
Hi @sdminonne , |
|
Can one of the admins verify that this patch is reasonable to test? (reply "ok to test", or if you trust the user, reply "add to whitelist") If this message is too spammy, please complain to ixdy. |
|
@lhuard1A sure. Let you know ASAP. |
|
Works here. Thanks! |
| @@ -80,7 +80,9 @@ function create-kubeconfig() { | |||
| fi | |||
|
|
|||
| "${kubectl}" config set-cluster "${CONTEXT}" "${cluster_args[@]}" | |||
| "${kubectl}" config set-credentials "${CONTEXT}" "${user_args[@]}" | |||
| if [ -n "${user_args[@]:-}" ]; then | |||
There was a problem hiding this comment.
Nit: per our chosen bash style guide [[ is prefered over [
https://github.com/GoogleCloudPlatform/kubernetes/blob/master/docs/devel/coding-conventions.md
There was a problem hiding this comment.
The rest of the file does 2 space indentation as well.
There was a problem hiding this comment.
in my limited testing, I think using [ instead of [[ is what broke this for GCE, though I can't entirely explain why.
There was a problem hiding this comment.
These actually evaluate differently. The [[ -n does the right thing with array expansion, whereas [ seems to, well, deal poorly. I love bash:
zml@zml:~$ args=("--foo" "--bar" "--blatz")
zml@zml:~$ if [[ -n "${args[@]:-}" ]]; then echo "WUMPUS"; fi
WUMPUS
zml@zml:~$ if [ -n "${args[@]:-}" ]; then echo "WUMPUS"; fi
-bash: [: too many argumentsThere was a problem hiding this comment.
that's what I discovered in my testing, too, but I didn't see that error in the build log on Jenkins, so I was/am confused.
There was a problem hiding this comment.
wait no, I'm just bad at searching the log:
/jenkins-master-data/jobs/kubernetes-e2e-gce-parallel/workspace/kubernetes/hack/e2e-internal/../../cluster/../cluster/gce/../../cluster/common.sh: line 83: [: too many arguments
|
LGTM |
mikedanese
added
the
lgtm
|
All green, and isolated change |
|
Pretty sure this change broke Jenkins for GCE: This change also didn't run through the per-PR Jenkins before being merged, making it even more suspect. |
|
Oh man, I missed that. I apologize. Shame on me. |
KUBERNETES=libvirt-coreos cluster/kube-up.shproduced the following error:This was coming from the fact that, as a
libvirt-coreoscluster runs locally on local VMs,there is no authentication mechanism. This led to have
user_argsofcommon.shunset.In the case of libvirt-coreos, it is in fact expected to have no authentication token.