Pipeline fails

[Brad Vrabete]

The pipe configuration below fails:

---

             - echo "Kustomize the deployment"

            - pipe: atlassian/aws-eks-kubectl-run:2.1.0 # runs kubectl apply -k against the specific folder; basically patching the initial files in base directory

              variables:

                AWS_ACCESS_KEY_ID: "$AWS_DEV_ACCESS_KEY_ID"

                AWS_SECRET_ACCESS_KEY: "$AWS_DEV_SECRET_ACCESS_KEY"

                AWS_DEFAULT_REGION: "$AWS_DEV_DEFAULT_REGION"

                CLUSTER_NAME: "$AWS_CLUSTER_NAME"

                KUBECTL_COMMAND: "apply"

                KUBECTL_ARGS: 

                  - "--dry-run=server"

                RESOURCE_PATH: "k8s/deploy/aws-dev"

                KUBECTL_APPLY_ARGS: "-k"

                LABELS: 

                  - "test.label=Test-deployment"

                WITH_DEFAULT_LABELS: "False"

                DEBUG: "True"

---

The error is:

INFO: Successfully updated the kube config.

Traceback (most recent call last):

File "/pipe.py", line 47, in <module>

pipe.run()

File "/root/.local/lib/python3.8/site-packages/kubectl_run/pipe.py", line 157, in run

self.handle_apply()

File "/root/.local/lib/python3.8/site-packages/kubectl_run/pipe.py", line 114, in handle_apply

self.update_labels_in_metadata(template_file, labels)

File "/root/.local/lib/python3.8/site-packages/kubectl_run/pipe.py", line 37, in update_labels_in_metadata

yaml_doc['metadata'].setdefault('labels', {}).update(labels)

TypeError: 'NoneType' object is not iterable

---

A label is defined so this should work. What can I do to fix it?

[Igor Stoyanov]

@Brad Vrabete Hi. Thanks for your question.  This error can occur only if `LABELS` is empty. We will try investigate this case and give you feedback soon.


Regards, Igor.

[Brad Vrabete]

Thanks, Igor. I tried to disable the labels as the branch name was longer than 63 characters accepted by Kubectl for a label. 

Now, if you can see the script, LABELS is definitely not empty:

LABELS: 

                  - "test.label=Test-deployment"

So how would label need to be defined?

[Igor Stoyanov]

@Brad Vrabete  Seems like you doing it right and the problem on our side. 

The only one temporary solution i can provide is to pass labels directly to your template file "k8s/deploy/aws-dev" under `kind: Deployment metadata`.
Example of the template:

template = """
apiVersion: apps/v1
kind: Deployment
metadata:
  name: test
  <pass your custom labels here>
spec:
  replicas: 2
  selector:
    matchLabels:
      app: nginx
  template:
    metadata:
      labels:
        app: nginx
    spec:
      containers:
      - image: nginx:1.7.9
        name: nginx
        ports:
        - containerPort: 80
"""

[Brad Vrabete]

I actually found an easier way: 

  script:

            # shrink bitbucket branch name to less than 63 characters

            - branch="${BITBUCKET_BRANCH}"

            - export BITBUCKET_BRANCH=${branch:0:62}

            - pipe: atlassian/aws-eks-kubectl-run:2.1.0 # runs kubectl apply -k against the specific folder; basically patching the initial files in base directory

              variables:

                AWS_ACCESS_KEY_ID: "$AWS_DEV_ACCESS_KEY_ID"

                AWS_SECRET_ACCESS_KEY: "$AWS_DEV_SECRET_ACCESS_KEY"

                AWS_DEFAULT_REGION: "$AWS_DEV_DEFAULT_REGION"

                CLUSTER_NAME: "$AWS_DEV_CLUSTER_NAME"

                KUBECTL_COMMAND: "apply"

                KUBECTL_ARGS: 

                  - "--dry-run=server"

                RESOURCE_PATH: "k8s/deploy/aws-dev"

                KUBECTL_APPLY_ARGS: "-k"

                WITH_DEFAULT_LABELS: "True"

[Igor Stoyanov]

Good to know you found a solution for your case. Anyway, thanks for your feedback about problem with LABELS you have found.

[Igor Stoyanov]

Hi @Brad Vrabete . The issue with LABELS fixed in aws-eks-kubectl-run:2.3.0 pipe .

Regards, Igor