Notes from the Ethereum CLC meeting today 🪡
FYI, there were no updates on Merge or difficulty bomb timing. But there were some cool client updates.
First up on the call, devs discussed how the 3rd shadow fork of Ethereum mainnet went.
Conversation
Replying to
If you don’t know what a shadow fork is, read this: github.com/timbeiko/eth-r. gave a rundown on the call of how today’s mainnet SF went.
1
3
8
Network participation was healthy. 99.8% of validators were online and active before the Merge but sometimes Prysm/Besu validators would miss blocks. After the Merge, participation dropped slightly to 97.6%. The Prysm/Besu validators that we’re having issues dropped off and
1
4
21
Prysm/Nethermind validators also had some issues with the transition. The issues are being investigated and/or fixed. With today’s SF, devs also purposefully made a few nodes fall out of sync leading up the Merge to see how they would handle resyncing to the chain post-Merge.
1
3
13
There were issues that devs are currently analyzing. More on that here: notes.ethereum.org/@parithosh/Hyi. Another SF for Merge testing will be scheduled for next week.
1
3
11
Devs reconfirmed what was discussed about difficulty bomb timing during last Friday’s ACD call. My recap of the call here:
Quote Tweet
Recap of the Ethereum All Core Developers’ Call today, starting with the topic of difficulty bomb timing.
youtube.com/watch?v=SWWoniShow this thread
1
4
15
Tldr, if all goes well with Merge testing in the next few weeks, devs may not have to delay the difficulty bomb. However, for that to happen, they would need to decide on dates this month for forking (for real) Ethereum testnets Ropsten, Goerli, and Sepolia.
5
14
51
Client teams are planning on only maintaining the latter two long-term and letting Ropsten either be deprecated after the Merge or given over to another entity to maintain.
1
4
22
. explained that in lead up to Merge activation on these 3 testnets, devs would have to spin up 2 new Beacon Chain testnets for the Ropsten and Sepolia PoW chains.
1
2
5
. stressed that the Beacon Chain testnets should ideally be spun up in the next 2-3 weeks so that they can be running for a good amount of time before devs activate the Merge on Ropsten and Sepolia. (Goerli already has a test Beacon Chain. It’s called Peermont.)
1
3
7
The plan is to make Sepolia a permissioned testnet unlike the others for Merge activation where only known entities can run validators and execute txs on the network.
1
3
5
This is so that devs can test out more chaotic states where half the validator set goes down, there’s long inactivity leaks and non finalization periods, etc., without having to worry about impacting unknown dapp developers also using the chain for other testing purposes.
1
2
7
To that end, devs are planning to spin up a much larger validator set for Sepolia than Ropsten or Goerli. In preparation for Merge activation on testnets, there’s a new release forthcoming of CL specs for clients to implement that will include changes to stuff like:
1
2
7
engine api timeouts, invalid terminal block status, and error codes. Client teams are encouraged to review the open commits on GitHub and start implementing the ones already merged! Moving on now to client updates:
1
1
6
- Lodestar team has implemented their version of proposed boost which safeguards Ethereum’s PoS consensus protocol from certain reorg attacks. More info on that here: arxiv.org/abs/2110.10086#. I think almost all of the client teams have implemented proposer boost now.
1
1
5
- Lighthouse is working on 2 cool projects. One to enable better support for CL nodes using the latest internet communication protocol known as IPv6. Likely won’t be ready for the Merge but it’s something devs are working towards. Lighthouse team is also working on
1
3
upgrading the CL gossipsub protocol which handles message propagation between nodes. More info on gossipsub here: arxiv.org/abs/2007.02754. After talking with one of the authors of this paper, believes there could be a way to make a backwards compatible change to
1
3
gossipsub that would greatly reduce the bandwidth requirements for CL nodes. It’s an ongoing area of research and Manning said he’d write to more documentation on it for interested client team in the upcoming weeks.
- Nimbus is working on supporting distributed validator tech.
1
5
I wrote about DVT in this thread here. Basically it enables multiple remote signers to combine their signatures for performing validator responsibilities.
Quote Tweet
Distributed validator technology (DVT) was a major topic of discussion today at the #StakingGathering event. DVT is the middleware tech that enables Eth validator responsibilities to be split up across multiple node operators.
Show this thread
2
6
Since no single node operator would have access to a full validator key, this would greatly reduce potential for malicious signing behavior by rogue actors in a staking pool. Nimbus is looking into how to better support such a configuration through their key manager API.
1
1
6
Some more info on what is currently supported by a few clients already for multiple remote signers: docs.web3signer.consensys.net/en/latest/
- Teku team is working on implementing the Builder API,
1
3
which will mainly be used by folks like Flashbots for creating the middleware to allow validators to outsource the task of block production to third part block builders. More info on Builder API here:
1
3
More info on the middleware called MEV Boost here: github.com/flashbots/mev-. This brings me to the last topic on the call I want to make note of. said that the specs for the Builder API should be ready for implementation and testing in CL clients in the next 2 weeks.
1
3
The way it works now it’s not possible for validators to compare the profitability of a locally created block and a block submitted by an external builder.
2
2
This functionality should be enabled sometime after the Merge through a new API query (get_payload_v2) so that validators can check there’s no nefarious behavior in terms of fees with externally received blocks versus locally built ones.
1
4
Until it’s created, there is some amount of trust validators need to put in the relays that route externally built blocks.
1
4
. highlighted that sentries can be set up to monitor such behavior and if there’s a lot of it post Merge devs can expedite pushing out the new API query.
1
4
And that’s a wrap for the CLC call today. Highly recommend checking out summary of the call too: hackmd.io/@benjaminion/H. Upwards and onwards! 🚀
2
1
14