All pods from helm test should end in a "Completed" state and not an "Error" state.

As the installer of Watson Assistant on Cloud Pak for Data, seeing anything that says "Error" is concerning and encourages me to open up a support case with IBM. A pod ending up in an "Error" status suggests that something has gone wrong in the test, even if the test says that it passed.

The test for watson-assistant-etcd3-auth-test should be modified so that it finishes with a "Completed" status (as all other pods do) rather than an "Error" status. The current behavior is below.

The test says that it passed:

RUNNING: watson-assistant-etcd3-auth-test
PASSED: watson-assistant-etcd3-auth-test

The created test pod is in an error state:

[root@virtualserver01 px-install-4.x]#‌ oc get pod watson-assistant-etcd3-auth-test
NAME READY STATUS RESTARTS AGE
watson-assistant-etcd3-auth-test 0/1 Error 0 38m

The logs for the test pod:

[root@virtualserver01 px-install-4.x]#‌ oc logs watson-assistant-etcd3-auth-test
watson-assistant-etcd3.zen.svc.cluster.local:2379 is unhealthy: failed to connect: etcdserver: authentication failed, invalid user ID or password
Error: unhealthy cluster

From describing the pod, the error in the logs makes sense because it is explicitly setting the wrong password in the environment:

 Environment:
 ETCDCTL_API: 3
 ETCDCTL_DIAL_TIMEOUT: 3s
 USERNAME: <set to the key 'username' in secret 'watson-assistant-assistant-etcd-creds'> Optional: false
 PASSWORD: <set to the key 'password' in secret 'watson-assistant-assistant-etcd-creds'> Optional: false
 ETCDCTL_USER: $(USERNAME):$(PASSWORD)-wrongPassword
 ETCD_CERT_FILE: /var/etcd/certs/tls.crt
 ETCD_KEY_FILE: /var/etcd/certs/tls.key
 ETCDCTL_CACERT: /var/etcd/certs/tls.cacrt

This change requires minimal effort and would have positive benefit to all users of Watson Assistant on Cloud Pak for Data. IBM would save money because our support organization will not have to respond to customer queries about non-existent errors.