Centralization Risks and Mitigation

Outlining potential centralization risks and their mitigations

Risk: Obol hosting the relay infrastructure

Mitigation: Self-host a relay.

One of the risks associated with Obol hosting the LibP2P relays infrastructure allowing peer discovery is that if Obol-hosted relays go down, peers won't be able to discover each other and perform the DKG. To mitigate this risk, external organizations and node operators can consider self-hosting a relay. This way, if Obol's relays go down, the clusters can still operate through other relays in the network. Ensure that all nodes in the cluster use the same relays, or they will not be able to find each other if they are connected to different relays.

The following non-Obol entities run relays that you can consider adding to your cluster (you can have more than one per cluster, see the --p2p-relays flag of charon run):

Entity	Relay URL
DSRV	https://charon-relay.dsrvlabs.dev
Infstones	https://obol-relay.infstones.com/
Hashquark	https://relay-2.prod-relay.721.land/
Figment	https://relay-1.obol.figment.io/
Node Guardians	https://obol-relay.nodeguardians.io/

Risk: Obol being able to update Charon code

Mitigation: Pin specific docker versions or compile from source on a trusted commit.

Another risk associated with Obol is the Labs team having the ability to update the Charon code used by node operators within DV clusters, which could introduce vulnerabilities or malicious code. To mitigate this risk, operators can consider pinning specific versions or hashes of the Docker image or git repo commits that have been thoroughly tested and accepted by the network. This would ensure that any updates are carefully vetted and reviewed by the community, and only introduced into a running cluster gradually. The labs team will strive to communicate the security or operational impact any Charon update entails, giving operators the chance to decide whether they want potential performance or quality of experience improvements, or whether they remain on a trusted version for longer.

Risk: Obol hosting the DV Launchpad

Mitigation: Use create cluster or create dkg locally and distribute the files manually.

Hosting the first Charon frontend, the DV Launchpad, on a centralized server could create a single point of failure, as users would have to rely on Obol's server to access the protocol. This could limit the decentralization of the protocol and could make it vulnerable to attacks or downtime. Obol hosting the launchpad on a decentralized network, such as IPFS would be a first step but not enough. This is why the Charon code is source-available and contains a CLI interface to interact with the protocol locally.

To mitigate the risk of launchpad failure, consider using the create cluster or create dkg commands locally and distributing the key shares files manually.

Risk: Obol custodying pre-signed exit messages

Mitigation: Use withdrawal address initiated exits or validator client exits

Before the Pectra hardfork, there was no way for a delegator to exit their validator without the co-operation of the node operator(s). In an effort to reduce this risk, Obol developed the Charon exit command, which allows operators to pre-sign and download exit messages to give to the delegator for safe keeping, ensuring they could broadcast them at any point in time. This feature relies on Obol's API, and means the Obol core team could in theory initiate an unwanted exit. If a delegator does not want to be exposed to that centralization risk, they should not use Charon's built-in exit commands, and instead should initiate exits using the EIP7002 exit contract, or alternatively by running normal exit commands on the operator's validator clients.

Guides to exiting validators using all three approaches are outlined here.

Risk: Obol going bust/rogue

Mitigation: Use key recovery.

The final centralization risk associated with Obol is the possibility of the company going bankrupt or acting maliciously, which could lead to a loss of confidence in the Charon client. To mitigate this risk, Obol has implemented a key reconstitution mechanism. This would allow the clusters to continue operating and to (re)create full validator private keys suitable for standard staking setups even if Obol is no longer able or willing to provide support.

A guide to recombine key shares into a single private key can be accessed here.