Has the option to set a docker host URL been removed?

[zero-ninetynine]

I've previously tried arcane before the 1.0 release and there used to be an option to set the URI for the docker host during onboarding, so I could use the TCP URL to my docker socket proxy and not have to expose the docker socket directly to arcane. This old closed issue refers to the same feature. I was excited to try out the arcane post-1.0 and just installed the latest version of arcane from scratch, but I can't for the love of me find any such option anymore. Was the option to use a socket proxy removed or is it somewhere I haven't found yet?

[spiralham]

Same, I can no longer figure out how to run arcane with my docker socket proxy.

[scottjamesrogers]

Same. In my case I'm trying to run the arcane-headless:v1.2.2 image on remote servers and connect it to tecnativa/docker-socket-proxy:v0.4.1 via Docker networking. While setting up the network is straightforward, it seems arcane-headless only supports interacting with the Docker API through a mounted Docker socket. Is there a way to configure it to use the proxy instead?

[makinikm]

In current versions of Arcane (2.15.0 as of now), there is no UI option for adding an Environment "agent-less" - using a URI (Docker\Podman API over TCP), be it local or remote. It is possible only via Arcane's Docker-proxy setup (without using a proxy), by setting the env DOCKER_HOST=tcp://docker-api:2375. The Arcane instance then only reflects that socket and it's host, treats it as the local Environment, ignoring the default Docker socket and running host.

If such option used to exist, I wonder if there was a specific reason it was removed... ?
It is quite useful in local or lab-like setups: local or remote environment(s) that don't require a separate 24/7 running agent, e.g. the "remote" URIs are virtual machines running on the same host. Running an Arcane agent in each will definitely cause excessive compute. Although an edge-case, there used to be that other CE project that is no longer developed, they had such option for Docker and never completed it for Podman URIs (but since both APIs are standard-compliant, those were interchangeable).

Could this perhaps be considered as a feature for resurrection?