HackMDWhat's New in Eth2 - 27 August 2021
- The image file may be corrupted
- The server hosting the image is unavailable
- The image path is incorrect
- The image format is not supported
Edition 76 at eth2.news
Public Service Announcements
See this Reddit post for more info and Geth upgrade instructions. If you don't update, you will likely become unable to propose beacon blocks. Also, Erigon nodes older than v2021.08.04 must be updated.
On PSA-2, here are the Prater–Altair supporting client versions you need:
- Teku: v21.8.2
- Lighthouse: v1.5.1
- Nimbus: Altair-beta (Do not run this version on mainnet!)
- Prysm: v2.0.0-rc.1 (Do not run this version on mainnet!)
The Beacon Chain
On Friday the 20th we had a very minor incident on the Beacon Chain. It's only the second interesting incident in the chain's nine month life so far. Honestly, if the chain didn't run near-perfectly most of the time, I doubt we would even have noticed anything was up.
The summary is that one of Lido's providers, that manages around 4000 validators (~2% of the total network), was suffering degraded service and producing a lot of orphaned blocks. This reduced the total block space available for including attestations, so a small number of attestations was getting dropped and not included on chain. The result was some (very) minor penalties for validators, and a drop in overall participation rate of a percent or two. Nothing dramatic. On further study it turned out that the problem had been amplified by (1) Prysm not handling the orphaned block situation correctly, and (2) Prysm also being the dominant client on the network.
Check the following write-ups for more info:
- My initial tweet thread explainer. Don't miss the fork.
- Lido's own Postmortem with lots of info about the root cause on their side. [Aside: why do we call it a postmortem? Nothing died.]
- A detailed data analysis of the incident from Shyam Sridhar of the Ethereum Foundation's Robust Incentives Group
In all other respects, things are going swimmingly!
Altair
The Altair beacon chain upgrade will be deployed on the Prater testnet on September 2nd, at epoch 36660. See PSA-2, above.
The date for upgrading the beacon chain proper has not yet been set in stone, but is likely to be September 30th, all being well. All stakers will need to update their clients by then, so get ready. You all keep them up to date anyway, right? Right?
The Altair upgrade on the Pyrmont testnet went pretty well. We are now using Pyrmont as a playground for stress-testing clients and the protocol. First we turned off half of the validators (client teams control most of the validators on Pyrmont) for three days to test behaviour during a long period of non-finalisation. All was fine, bar a syncing issue discovered in Prysm.
Danny's Finalized no. 28 outlines the road to Altair.
The Merge
There was agreement on the August 20th All Core Devs call that there will be no new EIPs included in Eth1 before The Merge. Getting to proof of stake is now the top priority for Ethereum devs :tada:
Dmitry Shmatko of ConsenSys TX/RX has analysed the behaviour of the transition total difficulty around The Merge. He makes some recommendations to help plan the timing of The Merge more precisely. Since hashrate in PoW is unpredictable, there is a natural variation to take account of.
Mikhail Kalinin, also of TX/RX, has proposed an API design for the interface between consensus clients and exection clients post-merge. This is being thrashed out among devs across a number of calls.
Staking
Open your ports, people! Adrian Sutton explains why open ports are important both for the health of your node and the health of the network.
The Great Explainers
I promised you videos about the formal verification of the beacon chain specification using the Dafny framework. And here they are!
Light clients will be an integral part of Ethereum 2.0 once it is fully formed. If you'd like to get familiar this technology you could do a lot worse than begin with Lodestar's light client explainer. Lodestar has built a prototype light client for the beacon chain that runs in browser.
Tomasz Stańczak of Nethermind and Flashbots did a PEEPanEIP session on 🤖 MEV & Flashbots 🤖. The second part is all about MEV in Eth2, post-merge. These PEEPanEIP sessions are quietly becoming an incredible resource.
Media and other stuff
I think I may have missed a shout-out to the Bankless Staking Panel when it first happened. In any case, I've listened to it now, and it's definitely worth a repeat. Fascinating to hear the perspectives from three very different staking operations. And, in truth, Darren Langley's voice is so trust-inducing that I now want to give him all my money, no questions asked. Audio here.
Superphiz's State of the Stake is back after a bit of a hiatus. Phiz updates us on the progress of the beacon chain. There's a lot packed into a short 11 minutes.
EthStaker Community Call #18 featured Saulius Grigaitis talking about the new Grandine client. Great discussion about Grandine's particular approach, and also why it remains closed source for now.
If you can't get enough EthStaker (trust me, you can't), here's Community Update 15 with Unvetica.
Christine Kim (who is leaving CoinDesk 😢) discusses a measure of decentralisation of staking that is trending in the right direction. This is based on some work by Nansen on The Data Behind Ethereum 2.0.
Research
Committee-driven MEV smoothing. MEV gains are very volatile by their nature. This proposal shares rewards from MEV with the whole set of validators attesting to a block at a slot. Overall it ought to help individual stakers not to be at a disadvantage to large pools that can smooth their MEV gains naturally. I'm not sure I like the feature whereby the extracted value is forced to be maximised - not every desirable outcome is measured in Gwei. But I am very happy to see continued innovation chipping away at the MEV world and its consequences. There is a longer write-up.
Exit/entry queue clogging after withdrawals are enabled. This covers two issues: partial balance withdrawals (or transfers) to allow staking rewards to be compounded; and, signing key rotation for validators. In the present design, both of these can be achieved only by exiting and re-staking validators, which is cumbersome and inefficient. Neither proposal is new, but it's good to have the discussion in one place to focus minds.
Regular Calls
Implementers
Call #71 took place on the 26th of August.
- Agenda
- Video
- My quick notes, and tweet-form from Alex Stokes.
In addition to the usual client team updates, we covered Altair planning, and discussed clients changing their PeerIDs on the network. Lighthouse and Teku tend to persist the beacon node's PeerID, whereas Nimbus changes it at every restart. We talk about the pro and con. Jacek expands on Nimbus's approach in this short thread.
API design for The Merge
Just ahead of the implementers' call, we met to discuss the design of the consensus API for The Merge. This is the new interface to be created between existing Eth1/execution clients and Eth2/consensus clients. The goal was to review the currently proposed API design.
Progress was slow; getting APIs right is hard, and getting them wrong can haunt you forever. Lots of good points were made. Discussion will continue on further calls and the Ethereum All Core Devs call.
StakeHouse Community Call
The StakeHouse community call has now become a fixture. Call #6 took place on the 25th of August, and will henceforth be happening at 3pm GMT every other Wednesday.
Contains impressive demos of Stereum, which is now a pretty complete GUI client installer and switcher, and the SSV.network testnet set up.
(Superphiz is excited by Stereum!)
Upcoming events
- EDCON is on right now! There is a bunch of Eth2 talks and familiar faces. Check out the agenda.
- Saturday, August 28th @ 10:00pm UTC: EthStaker is co-hosting a livestreamed workshop with Rocket Pool on setting up a node. More agenda info here.
- Monday, August 30th @ 11:00am UTC: SSV AMA #2 with Blox staking. There will be POAPs! (SSV is secret shared validators.)
In other news
- The ssv.network testnet is live and looking for people to try it out.
And finally…
This edition brought to you with the help of 1969 Reuben Wilson.
Learn More →
Learn More →
Advertising on this newsletter.
