Fix panic if maxreplicas is set but autoscale is not defined in the CR

[kevinearls]

Signed-off-by: Kevin Earls kearls@redhat.com

Fixes #1197

[jsirianni]

Thanks for handling this so fast. I built the image and tested it against a development instance. It does seem to resolve the panic, but I do not see an HPA anymore. In the past, a horizontal pod autoscaler would be deployed.

It is unclear to me if the HPA issue is related, or a separate issue.

[kevinearls]

@jsirianni what does your CR look like? I used the one below and that worked for me. You may need to first make sure you've deleted any previous collector instances. Use kubectl get otelcols --all-namespaces=true to find them.

apiVersion: opentelemetry.io/v1alpha1
kind: OpenTelemetryCollector
metadata:
  name: simplest
spec:
  minReplicas: 1
  maxReplicas: 2
  config: |
    receivers:
      jaeger:
        protocols:
          grpc:
      otlp:
        protocols:
          grpc:
          http:
    processors:

    exporters:
      logging:

    service:
      pipelines:
        traces:
          receivers: [jaeger,otlp]
          processors: []
          exporters: [logging]

[jsirianni]

@kevinearls you were right, looks good on my end now 👍

Thanks again

[frzifus]

Is it covered by an e2e test?

[kevinearls]

@frzifus This should be covered by the unit test. We just need to be sure that the AutoScalerSpec is not null when the CR has maxreplicas set but does not include autoscale.