Currently it is not easy to include the use max-nodes-totalcli option within any metrics, except for putting it there manually.
It would be nice if the metrics endpoint also exposes the value passed to max-nodes-total as it could then easily be queried in prometheus e.g. to generate alerts if the amount of nodes is coming close to that value?
Why is the manual configuration not enough? max-nodes-total is used as a safe-guard from the cluster to go crazy and provision too many resources. But sometimes the newly demanded resources are really needed and one will tune that setting. Meaning it always need to be set in two places.
although this metric would be static per deployment of the autoscaler, i think having this would also help empower users to create more alerts around this value too. thanks for the suggestion!
/kind feature
It should be fairly easy to add this and I'm happy to help review/etc.
/assign
Exposed via #3435 as cluster_autoscaler_max_nodes_count.
Most helpful comment
It should be fairly easy to add this and I'm happy to help review/etc.