Relay node permissionless-ness clarification

There has been some misunderstanding on the nature of the relay node whitelist that I would like clarified if possible:

My understanding is that the whitelisted relays by the foundation are the only ones propagated to clients (either via DNS SRV record or gossip protocol).

Private, non-permissioned relays can join the network and accept “relay traffic” but do not get advertised to clients. To use them you have to explicitly set them up as a peer on algod config.

Is this correct?

This is not a criticism of the permissioned state of relays in the least btw - just asking a clarification due to my initial misinterpretation of this reddit post: Running a relay is not permissioned. Getting paid to run one is. : algorand

1 Like

This is correct.

See General Frequently Asked Questions | Algorand Foundation

In particular:

Q16: Who is running relay nodes on the Algorand network?

While in principle anyone can run a relay node, the default behavior of an Algorand node is to only connect to relay nodes from a list that the Algorand Foundation maintains.

[…]

Q18: Who manages the list of relay nodes? What about decentralization?

Currently, the Algorand Foundation manages the official list of relay nodes, to bootstrap a scalable and reliable initial infrastructure backbone.

It is important to stress, however, that anybody with an Algorand account can run a non-relay node and participate in the Algorand consensus protocol (i.e., be a validator).

Moreover, the integrity of the blockchain does not depend on the relay nodes: as long as sufficiently many participation nodes (in terms of stake) behave honestly, the blockchain cannot fork.

Even if all the relays misbehave, the worst that can happen is that the blockchain will slow down or stall.

The Algorand Foundation is researching options for making the decisions on relay nodes in a more decentralized way.

1 Like