RPL Node Balancing

Node Operator Experience
1

Leaving the withdrawal of RPL under 150% as a separate topic, this is a request to remedy an issue for node operators managing more than one node.

As part of a risk management strategy, it became prudent to put a cap on the number of minipools/validators on a single machine. This could be to minimize risk of downtime or minimize risk of a compromised validators key. As Node 1 was stood up first, it received the bulk of the RPL Collateral. Once the need for Node 2 was identified, there was not enough RPL to sufficiently collateralize the node. Node 1 remains in excess (but still under 150%) while Node 2 remains deficient.

I am requesting an ability to verify ownership of more than 1 node operator address, and upon verification we unlock an ability to redistribute RPL from 1 node to the other. This ensures no RPL is withdrawn.

As a direct reference of how this improves the Rocket Pool protocol, if this functionality were enabled, my client would be able to distribute enough RPL from Node 1 > Node 2 so that Node 2 has enough RPL collateral to support an additional 175 LEB8 Minipools (1400 ETH). Since this is not available to us, 1400 ETH will instead be moved to a machine dedicated to non Rocket Pool solo validators.

1 Like
2

Would your client be open to a method of running a single node on multiple machines? HA using Dirk/Vouch or DVT would both provide that ability.

1 Like
3

Yes, operating with a DVT architecture is on the roadmap but not something we’ve developed or tested yet. I suppose this increases its priority.

However, that wouldn’t reduce risk on any validator key concerns. It also creates a situation where we’ve already diluted the collateral by standing up node 2. A DVT architecture would allow me to go above my soft-cap of 300 minipools per validator key, but still not the ideal solution. Aggregating RPL over the entirety of the architecture allows the assumed calculations (ETH vs RPL) to stay intact.