Proposal / RFE

The cilium-operator (soon to be one IPAM binary per cloud integration) should have its health check take into consideration if it can successfully communicate and take actions against the cloud API.

Is your feature request related to a problem?
In my specific case I deployed the operator with some misconfigured AWS creds (using the AWS ENI IPAM) and the operator always reported healthy even though all AWS EC2 API calls were failing.

Describe the solution you'd like
Talking with @tgraf and @errordeveloper in slack there are a couple of options:

• do a one time check at the startup of the operator if it can do all the actions it will need through its lifetime
• fail the health check if X consecutive API calls fail in a row

I'm happy with both -- just need to define clearly on what conditions the API calls would fail.