The AI Model Management sample Server is deploying.

  Release:   {{ .Release.Name }}
  Namespace: {{ .Release.Namespace }}
  Endpoint:  opc.tcp://{{ include "ai-model-management.fullname" . }}.{{ .Release.Namespace }}.svc:{{ .Values.service.port }}/ModelManagementServer

Reach it from your machine:

  kubectl --namespace {{ .Release.Namespace }} port-forward svc/{{ include "ai-model-management.fullname" . }} {{ .Values.service.port }}:{{ .Values.service.port }}
  dotnet run --project samples/AI/ModelManagementClient -- opc.tcp://localhost:{{ .Values.service.port }}/ModelManagementServer

What this Server will tell a client about itself:

  Inference runs   {{ .Values.backend.site }}
  Jurisdiction     {{ .Values.backend.dataJurisdiction }}
  Egress permitted {{ .Values.backend.egressPermitted }}
  Input retained   {{ .Values.backend.retainsInput }}
{{- if .Values.ai.enableFallback }}
  Fallback         published, reaching {{ .Values.fallbackBackend.endpointUri }}
{{- else }}
  Fallback         none; a failed inference fails
{{- end }}

{{- if not (include "ai-model-management.hasCredential" .) }}

No credential is mounted. That is correct for an anonymous endpoint or for
workload identity, and wrong for anything else.
{{- end }}
{{- if .Values.credentials.create }}

WARNING: credentials.create put the secret through Helm, so it is in the release
history and in `helm get values`. Use credentials.existingSecret for anything
that is not a local cluster.
{{- end }}
{{- if not .Values.persistence.enabled }}

WARNING: persistence is off, so the Server generates a new application instance
certificate on every restart and every client has to trust it again.
{{- end }}
