Hi,
I’m looking for a solution that “switches” the HTTP requests to an upgrading (or unavailable) service, to a “maintenance” service (i.e. a simple nginx service with a temporary service unavailable page).
Since the current version of Rancher LB (https://github.com/rancher/rancher/issues/8015, https://github.com/rancher/rancher/issues/2777) does not allows to correctly manage the zero downtime when upgrade a service with scale=1, with the effect that the LB responds with a “503 Service Unavailable” error, I’m looking for a solution (an LB configuration) that dynamically switches the traffic on a pre-defined maintanance service when an error occurs (service unavailable or other backend errors).
Thank you