{{- $cli := ternary "oc" "kubectl" (eq .Values.platform "openshift") }}
BlazeMeter agent "{{ include "bzm-opl.fullname" . }}" installed into namespace {{ .Release.Namespace }}.

  Location (harbor): {{ .Values.harborId }}
  Agent (ship):      {{ .Values.shipId }}
  Image:             {{ include "bzm-opl.craneImage" . }}
  AUTH_TOKEN:        {{ if .Values.useSecret }}Secret/{{ include "bzm-opl.secretName" . }}{{ else }}ConfigMap/{{ include "bzm-opl.configMapName" . }} (plain text){{ end }}

1. Watch it come up:

   {{ $cli }} -n {{ .Release.Namespace }} rollout status deploy/{{ include "bzm-opl.fullname" . }}
   {{ $cli }} -n {{ .Release.Namespace }} logs -l role={{ include "bzm-opl.roleName" . }} -f

2. Confirm the agent shows *online* in BlazeMeter under
   Settings -> Private Locations, then run a test against the location.

Engines need {{ include "bzm-opl.engineCpuLimit" . }} CPU + {{ include "bzm-opl.engineMemoryLimit" . }} RAM + ~60GB disk (40GB of it /tmp) per concurrent
engine, and egress to *.blazemeter.com{{ if .Values.privateRegistry }} and {{ .Values.privateRegistry }}{{ else }} and gcr.io{{ end }}.
Check a cluster against that before a real run: bzm-opl-gen doctor
{{- if and (not .Values.privateRegistry) (not .Values.proxy.enabled) }}

Note: with no privateRegistry and no proxy, the nodes pull images from the
public internet directly. In a locked-down cluster that is the first thing to
fail, and it fails as ImagePullBackOff on the *engine* pod mid-test, not here.
{{- end }}
{{- if not .Values.useSecret }}

WARNING: AUTH_TOKEN is in a ConfigMap, readable by anyone with get on
ConfigMaps in {{ .Release.Namespace }}. Reinstall with useSecret=true to move it.
{{- end }}
