aggregate label is problematic for hosts which belong to multiple aggregates

[Vultaire]

We've found an issue regarding metrics, observed concretely on the hypervisor_schedulable_instances metric but likely affecting others, where a metric only shows up under one of the aggregates that a host is a part of.

For example, if a host is part of aggregates a, b, and c, if I were to search for hypervisor_schedulable_instances{hypervisor_hostname="my-host.example.com"}, I might get output that looks like:

Key	Value
hypervisor_schedulable_instances{aggregate="a", arch="x86_64", cloud="my-cloud", hypervisor_hostname="my-host.example.com"}	175

That is - it would give me a single record indicating 175 schedulable instances on my host. However, it would only show the record under aggregate "a". If I were to change my query to sum/avg over aggregates "b" or "c", the above record wouldn't even be included in the calculation.

I'm not a Prometheus guru so I don't have a suggestion for how to correct this in a sane way; adding duplicate records with different aggregate labels would mess up sum()/avg() over the entire collection, but I am not sure of how else to correct this... Or should per-aggregate metrics be collected in some alternative way? I don't know; I leave that for you to review and consider.

Thank you.