kubectl rollout status not working anymore as expected

[Paolo Mainardi]

Hello! In the last few days we noticed that "kubectl rollout status" is not working as expected anymore.

An example:

k8s ❯ kubectl get deployments -napp
NAME                            DESIRED   CURRENT   UP-TO-DATE   AVAILABLE   AGE
site-develop-w071nb             1         1         1            0           2d


k8s ❯ kubectl rollout status -napp -w deployment/site-develop-w071nb
deployment "site-develop-w071nb" successfully rolled out

As you can see that deployment is not yet available but kubectl just ignore it, it is a behaviour cross-deployment and it started some days ago.

We are using GKE:

* Master version 1.6.6

* Default pool 1.6.2

We upgraded the master version to 1.6.6 some days ago and it could be the culprit.

Do you have any ideas ?

[Irina (Google Cloud Support)]

Hello Paolo, 

Is it possible that before your ran the second command the pod had become available?

I followed the same steps by using pointed GKE Master and Default Node Pool versions, and got the following kind of messages which are expected according to this link:

“

Waiting for rollout to finish: 1 out of 5 new replicas have been updated…

Waiting for rollout to finish: 2 out of 5 new replicas have been updated…

”

[Paolo Mainardi]

Hello Irina,

No, i am quite sure that it doesn't work anymore since we upgraded the master to 1.6.6, we have several CI jobs that now just exit in a few seconds, because the rollout 

return too early, anyway i've just tried again:

~ ❯ kubectl rollout status -n app -w deployment/site-develop-w071nb

deployment "site-develop-w071nb" successfully rolled out

~ ❯ kubectl get deployments -n app

NAME                            DESIRED   CURRENT   UP-TO-DATE   AVAILABLE   AGE

site-develop-w071nb             1         1         1            0           2d

[Paolo Mainardi]

Please follow this thread: https://groups.google.com/forum/#!topic/kubernetes-sig-apps/xC0jsXyECeM