@mrnicegyu11 Do we really want to show maintenance page in case webserver is down? I think that we better show maintenance page when we really do maintenance in which case we can always set up it explicitly. In case we show maintenance page every time the webserver cannot be reached. It can make users think that all fine and the maintenance is happening while we are actually have no clue that something is broken?

The public-api is a machine-to-machine API. A fallback for maintenance should respond with status 503

• status 503
• response JSON such as

{
  "errors": [ "Under maintenance ... (human readable message)" ]
}

• headers attributes (with Retry being optional)

content-type: application/json 
Retry-After: <http-date>
Retry-After: <delay-seconds>

After discussion with @sanderegg, we decided that this shall probably live in osparc-simcore repository (I am not sure why exactly so). To keep fallback service with main services in sync, we need to introduce ENV VARs for shared traefik configuration. In simcore we can also add integration tests to test fallback logic. In ops-env we will need to override configs definitions to ensure smooth config updates.

This is taking way more than expected. I will probably pause this for now.