Decentralized technologies have always been of great interest to me. In my opinion, the decentralization of the Internet was one of the keys to its success, but year after year, we are gradually losing it. This is happening because we increasingly rely on "mainstream" services for convenience, habit, or trend.

I often hear colleagues say, "I've put my client's email on Y (Y = one of the major players in the email world) to avoid problems." The problems remain, but the colleague can "pass the buck" to the big player of the moment. And, in my experience, clients tend to be much more forgiving when the problem is with a big player.

Once, people used to say: "Nobody ever got fired for buying IBM." Today, I would say, "Nobody ever got fired for putting email on [Google, Microsoft, etc.]."

A few days ago, a colleague mentioned that he struggles to find some of his friends and colleagues on the fediverse because they are spread across various instances, and a simple search doesn't always yield the desired results.

Own Your Data

The fediverse is an extremely effective, resilient, autonomous decentralized communication technology. Unlike email, the many software implementations (such as Mastodon, GoToSocial, Mitra, Akkoma, Pleroma, etc. - soon snac will also support this type of operation) support the ability to "move" one's account. This means that when a user decides to relocate, the server will notify all followers that the user has moved. The followers will automatically stop following the original account and start following the new one. As of today, it is not possible to move past content (which will still be available on the old server).

In my opinion, every government, public entity, association, foundation, etc., that needs to communicate with the public should have its own communication channel, with full control over its data and the messages it delivers. When I read "my Discord server," I feel like responding "there's nothing 'yours' about it - tomorrow morning they could shut everything down, and you would have lost EVERYTHING.". Own your data!

Sometimes, I read that instances are not opened because of "costs not balanced by the number of users." But even public television channels are often economically unprofitable, yet they are considered an essential service for public communication. Open, decentralized technologies that ensure control over one's data should be treated the same way.

The problem, as with email, is tied to the address. A server change necessarily implies an address change, which can sometimes cause problems.

A Permanent Address

For this reason, I decided to implement a kind of "permanent" address that will provide the correct data based on my main server. I decided that this address should be linked to my email address, and since I control the entire domain (including the web server), I can do this quite simply. In this way, if I change servers, users who already follow me will be moved by the "move" operation, while those who do not yet follow me will still be able to find me using my "permanent" address, which will, in turn, provide the new user on the new server.

My setup will be very similar to what I found in an interesting blog post that I based my approach on.

In Practice

• My email address is stefano@dragas.it, and I want anyone from any fediverse instance to be able to type it in and find me.
• My main account is currently on BSD Cafe: https://mastodon.bsd.cafe/@stefano.

To achieve this, I modified the nginx configuration on the server that serves the "dragas.it" website.

Nginx Configuration

Before the server directive for dragas.it, I added this mapping:

map $query_string $account_name {
    ~resource=acct:stefano@dragas.it$ stefano;
    ~resource=acct%3Astefano%40dragas.it$ stefano;
}

This ensures that requests for those two resources (some clients send the second one, so I wanted to handle that situation) will assign the value "stefano" to the $account_name variable.

WebFinger Exception

Within the server directive of the dragas.it virtual host, I added an exception for webfinger (the protocol used to find user data):

location = /.well-known/webfinger {
    root  /usr/local/www/webfinger;

    if ($account_name) {
      rewrite ^(.*)$ /$account_name.json break;
    }

    try_files $uri = 404;
}

JSON File for WebFinger

At this point, I created the /usr/local/www/webfinger directory and added a file named stefano.json (matching $account_name) with the following content:

{
  "subject": "acct:stefano@dragas.it",
  "aliases": [
    "https://mastodon.bsd.cafe/@stefano",
    "https://mastodon.bsd.cafe/users/stefano"
  ],
  "links": [
    {
      "rel": "http://webfinger.net/rel/profile-page",
      "type": "text/html",
      "href": "https://mastodon.bsd.cafe/@stefano"
    },
    {
      "rel": "self",
      "type": "application/activity+json",
      "href": "https://mastodon.bsd.cafe/users/stefano"
    },
    {
      "rel": "http://ostatus.org/schema/1.0/subscribe",
      "template": "https://mastodon.bsd.cafe/authorize_interaction?uri={uri}"
    }
  ]
}

In practice, this will return a reference to my main Mastodon account.

Testing the Setup

After reloading the nginx configurations, it will be enough to test it:

# curl "https://dragas.it/.well-known/webfinger?resource=acct:stefano@dragas.it"

{
  "subject": "acct:stefano@dragas.it",
  "aliases": [
    "https://mastodon.bsd.cafe/@stefano",
    "https://mastodon.bsd.cafe/users/stefano"
  ],
  "links": [
    {
      "rel": "http://webfinger.net/rel/profile-page",
      "type": "text/html",
      "href": "https://mastodon.bsd.cafe/@stefano"
    },
    {
      "rel": "self",
      "type": "application/activity+json",
      "href": "https://mastodon.bsd.cafe/users/stefano"
    },
    {
      "rel": "http://ostatus.org/schema/1.0/subscribe",
      "template": "https://mastodon.bsd.cafe/authorize_interaction?uri={uri}"
    }
  ]
}

In this case, everything works perfectly. When you search for my email address from any fediverse instance, my BSD Cafe profile will appear.

Conclusion

This approach allows you to define multiple users for each domain. You just need to add the corresponding mapping in the nginx configuration and the .json file with all the data.