We can not disable haproxy_backend_status because we need to wait for APIs to come up at some points during deployment. I think we should refactor haproxy_backend_status (and probably rename it to backend_status) so it could support checks via haproxy status (current implementation) and something like basic "http_check" (connect to endpoint and check if it returns HTTP/200 or something like that). In such case we would be able to switch to 'haproxy' check on default deployments and to 'http' check on deployments with external non-haproxy LBs.
We can not disable haproxy_ backend_ status because we need to wait for APIs to come up at some points during deployment. I think we should refactor haproxy_ backend_ status (and probably rename it to backend_status) so it could support checks via haproxy status (current implementation) and something like basic "http_check" (connect to endpoint and check if it returns HTTP/200 or something like that). In such case we would be able to switch to 'haproxy' check on default deployments and to 'http' check on deployments with external non-haproxy LBs.