Connectors NPE due to missing operate URL when identity disabled

[falko]

Expected: Connector configuration understands that identity is turned off via global.identity.auth.enabled: false and identity.enabled: false.

But this config map ends up without any URL for Operate:

camunda-platform-helm/charts/camunda-platform-8.6/templates/connectors/configmap.yaml

Lines 28 to 35 in 4aca7ec

	{{- if or (eq .Values.connectors.inbound.mode "credentials") (and .Values.global.identity.auth.enabled (eq .Values.connectors.inbound.mode "oauth") ) }}
	operate:
	client:
	url: {{ include "camundaPlatform.operateURL" . | quote }}
	{{- if eq .Values.connectors.inbound.mode "credentials" }}
	username: connectors
	{{- end }}
	{{- end }}

which causes a Nullpointer Exception in the Connector Runtime:

As a workaround, I set connectors.inbound.mode: disabled and that worked.

I guess, connectors.inbound.mode: credentials would also work and if so, that would my proposal for a fix.

[falko]

The customer use case for this type of installation is:

• Small test system for development and testing on AWS EKS
• K8s namespace has very limited resources
• Installation preformed by a first-time user Helm
• kubectl port-forwarding is sufficient to access Operate and deploy to the Zeebe Gateway via Desktop Modeler

[hamza-m-masood]

Hello!
@falko could you please specify the helm chart version?

[falko]

It should have been the latest version at the time of my posting.

[hamza-m-masood]

That would be 11.1.0

[falko]

Actually, the customer is partially air-gapped. So they downloaded the Helm chart sources as a zip file from GitHub around the time when I opened the issue and the pointed helm to install locally from the charts/camunda-platform-8.6 folder. I downloaded the dependencies using make helm.dependency-update and sent them via email.

[hamza-m-masood]

Thanks, this makes sense to me.

[p4tr0ckz]

I'm running into the same issue. Thats why i initially downgraded to Chart Version 9.2.0 and opened #2357
But now i need to update to latest version and this came back up like this issue.

@falko Did your customer find a solution?