I have the following service where I want to select a pod from a label that exists.
But it does not work, do you know why?
Note: mypod: exists in the followin:
apiVersion: v1
kind: Service
metadata:
name: myservice
labels:
run: myservice
spec:
ports:
- port: 8080
targetPort: 80
selector:
mypod: exists
type: LoadBalancerThe pod that I am trying to select is listed kubectl get pods --show-labels:
NAME READY STATUS RESTARTS AGE LABELS
mypod-5bf4c474c8-l6rr2 1/1 Running 0 32m mypod=49c8b466-2fdd-4ea6-b799-bc60c259a8ecI don't have controle over the uuid and I know there will be only one pod - that is why I am trying to use exists.
Also if you know, what happens if there is more than one?
But most important, why does it not find the pod with exists?
In addition to @Michael Hausenblas answer.
Kubernetes is using 2 types of labels selectors - Set-based and Equality-based.
Because the Service object is old and was created before Set-based requirements were released, you cannot use rules like if label 'myapp' exists, you should use equals like if label 'myapp' = 'abc'.
So, if your CI/CD is changing the value of myapp label, you can add some additional one with a static value, like appname: myapp and use it in the Service selectors. There are no other ways to do it.
As you can see from your kubectl command output, the value of the label with the key mypod is in fact 49c8b466-2fdd-4ea6-b799-bc60c259a8ec and not exists. So I suppose your CI/CD pipeline overwrites this? Anyways, a simple key-only query like kubectl get po -l=mypod should work.