Missing metrics - example: cluster_autoscaler_failed_scale_ups_total 

Always 

#curl the autoscalers metrics endpoint: 

$ oc exec deployment/cluster-autoscaler-default -- curl -s http://localhost:8085/metrics | grep cluster_autoscaler_failed_scale_ups_total 
    

the metrics does not return a value until an event has happened   

The metric counter should be initialized at start up providing a zero value

I have been through the file: 

https://raw.githubusercontent.com/openshift/kubernetes-autoscaler/master/cluster-autoscaler/metrics/metrics.go 

and checked off the metrics that do not appear when scraping the metrics endpoint straight after deployment. 

the following metrics are in metrics.go but are missing from the scrape

~~~
node_group_min_count
node_group_max_count
pending_node_deletions
errors_total
scaled_up_gpu_nodes_total
failed_scale_ups_total
failed_gpu_scale_ups_total
scaled_down_nodes_total
scaled_down_gpu_nodes_total
unremovable_nodes_count 
skipped_scale_events_count
~~~