Socket hangup when forcing restify client to retry ad infinitum

This is a cross-post from my question on SO. I suspect this is a configuration mistake on my end, and would appreciate any help understanding where that mistake lies.

I am running several Restify v4.2.0 services in separate processes. One of them is special because it needs to be up and listening before all of the others even start.

Assume that I cannot use IPC, and all services must talk to the special service via HTTP. This leaves me with HTTP clients that attempt to contact the special service ad infinitum, waiting until it is up.

Hence this client with intentionally extreme options.

const client = restify.createStringClient({
    url,
    retry: {
        retries: Infinity,
    },
});

// Inside a promise
client.get(endpoint, function(err, req, res, data) {
    if (err) {
        reject(err);
    } else {        
        // ...

        // Probably not needed for GET, but here anyway.
        req.end();
        resolve(data);
    }
});

No matter how I configure the client, I still immediately get socket hang up errors. I want the nodeback to never fire until it finally reaches the service.

What do I need to do to my configuration to accomplish this?

Start with the Restify v4.2.0 client configuration shown in the issue, especially createStringClient, the retry option, and the client.get callback. Reproduce the socket hang-up behavior with retries set to Infinity and compare it with the intended retry behavior. Done means identifying whether the configuration is supported and documenting or correcting the behavior.