Feature #21938
Updated by Peter Amstutz 5 days ago
* total vCPUs (since AWS quotas are based on vCPUs)
* "instance type creation errors" that is parameterized on cluster, instance type and subnet. Either cumulative or rolling count of errors with a period of last 5-10 minutes
** only want to log fatal errors where it couldn't start an instance despite trying every subnet / every candidate instance type