PoS Implementers’ Call #76 - 2021-11-18

--- tags: eth2devs description: Notes from the regular proof of stake [Eth2] implementers call image: https://benjaminion.xyz/f/favicon-96x96.png --- # PoS Implementers’ Call #76 - 2021-11-18 [Quick contemporaneous notes by Ben Edgington; fka "Eth2 Implementers' Call"] Agenda: https://github.com/ethereum/pm/issues/418 Livestream: https://youtu.be/31Jxh9_xXvY ## Kintsugi office hours ### merge-devnet-0 [Pari] devnet was started a couple of hours ago, hasn't hit the transition yet. [Configs are here](https://github.com/eth-clients/testnet-configs/tree/main/merge-testnets/merge-devnet-0). There is a deposit contract, and Pari has some extra validators available if you want one. There is an explorer, but it has some issues right now. It is likely that optimistic sync will fail on this network, so it may be hard to join for new people. Marius plans to mine post-merge and generally cause chaos. Composition: - EL: Geth, Nethermind. - CL: Prysm, Lodestar, Lighthouse, Nimbus Plan to stand up a new devnet in one week. ### updates The [milestone doc](https://notes.ethereum.org/@djrtwo/kintsugi-milestones) indicates that a few clients have hit milestone M2. Lighthouse has finished on-merge-block consensus tests. Working on milestone M3, but slow getting everything set up. Working on optimistic sync. Only lockstep sync running now. Prysm, similar to LH. Lockstep sync working with Geth, will try with Nethermind today. Working on optimistic sync. Lodestar, have integrated Geth and Nethermind in CI. Optimistic sync PR is up. Nimbus, reached M2 with Geth and Nethermind, but behind on optimistic sync. Teku, working in parallel on optimistic sync and engine API implementation. Geth, created a branch: "beacon on top of 4399" - this is the branch to use to join the merge testnet. Trying to get Amphora specs into the master branch, which is a delicate process. Looking good. That branch has normal sync until TTD, and after that changes to reverse header sync. Nethermind, have clients on M2 milestone: LH, Lodestar, Nimbus. Hopefully Prysm soon. Looking at integrating with mergemock. There is a nice test tool for execution layers from Mario at the EF. There is disagreement about how much the EL should check the timestamp. Q: What resolution should we use for checking the timestamp? [Lightclient] is working on a testing tool. Allows people to write full featured tests in Python. Supports both engine API and transitions of blocks. Mergemock is likely to be the runner for executing the tests, but equally Mario's test tool could do this. ### TTD override issues Péter S (Geth) does not like the mechanism in which the consensus client can override the TTD in the execution client as it means very intrusive changes. [Mikhail] has [a doc](https://github.com/ethereum/consensus-specs/issues/2724) outlining options: 1. No override for TTD on the command line. An emergency TTD change (or change to terminal block hash, TBH) would require client releases to update (for both EL and CL). This may delay any action due to time for operators to update nodes. It may be acceptable nonetheless. TBH override is more time sensitive than TTD. There is a danger that operators only upgrade one of the two clients, which would be a problem. 2. Set the overrides separately on the CLI for each client. This may be quicker to execute. 3. EL leads and CL follows - but this probably can't be made to work. Can do both 1 & 2 and give users the choice, but this doesn't improve on 1 only in terms of time to execute. Will continue to work on this issue and any spec changes [there](https://github.com/ethereum/consensus-specs/issues/2724). Update EIP-3675 first, and work on the UX second. ## Other client updates **Lighthouse** Working on Flashbots integration for post-merge Ethereum. Working on standard key manager API. Looking into an edge case in fork choice. ## Spec discussion and AOB [Issue 2727](https://github.com/ethereum/consensus-specs/pull/2727) raises a bug in fork choice. The solution is simple and simpler than the existing approach. Will be in the next spec release. Proposer boost fix will be rebased on top of this, and some more merge transition vectors added. Getting proposer boost done prior to or at the merge is on the critical path. [Jacek] [PR 2649](https://github.com/ethereum/consensus-specs/pull/2649), historical batches. There looks to be a simple way to introduce this. It's a step towards a world in which block data can be stored off-client and verified very elegantly. Would be good to get done prior to the Merge as it would be useful for handling execution blocks. [Danny] may be better to postpone to Shanghai upgrade (the first post-Merge upgrade), which is light on the consensus layer. [Zahoor] does it even make sense at all to store execution payloads on the consensus layer? [Jacek] deduplication has been discussed, e.g. just storing the hash on the consensus side. Separate issue is where we store the actual execution data long term. Geth already has some capability to insert off-line data and propagate it. The new proposals combine nicely to help with both the deduplication and out-of-band storage issues. [Saulius] Any discussion around moving deposit contract logic to the execution client? (That is, to implement a native way to handle deposits rather than using the contract.) [Danny] Most discussion is around just shortening the follow distance to make deposits quicker. Nothing around removing the contract altogether. [Pooja] PM repo changes. Where to put notes for these meetings now? Tim will make a folder. Do past meeting notes need to be migrated? Yes - that makes sense. [Leo] Concerning the crawlers. Nodewatch is now filtering peers by client version, but this is still not sufficient: multiple networks, not only the mainnet are continuing to show up. Have submitted a [PR](https://github.com/ChainSafe/eth2-crawler/pull/123) [Hsiao-Wei] How to name The Merge upgrade parts? Community vote? Should be named afer a "B*" star. Both the execution and consensus upgrades will need names since they will not be simultaneous. * * * # Chat highlights From danny to Everyone: 02:05 PM : https://github.com/ethereum/pm/issues/418 https://github.com/eth-clients/testnet-configs/tree/main/merge-testnets/merge-devnet-0 https://notes.ethereum.org/@djrtwo/kintsugi-milestones From parithosh to Everyone: 02:06 PM : Ah I forgot to share the composition: EL: Geth, nethermind. CL: Prysm, Lodestar,Lighthouse,Nimbus From danny to Everyone: 02:18 PM : https://github.com/ethereum/consensus-specs/issues/2724 From parithosh to Everyone: 02:29 PM : I’d set a much higher TTD next week and release configs a day earlier, so anyone can join in and go through the merge themselves. From danny to Everyone: 02:30 PM : https://github.com/ethereum/keymanager-APIs/ From Mehdi Zerouali to Everyone: 02:30 PM : https://hackmd.io/@paulhauner/H1XifIQ_t From Hsiao-Wei Wang to Everyone: 02:31 PM : https://github.com/ethereum/consensus-specs/pull/2727 From danny to Everyone: 02:33 PM : https://github.com/ethereum/consensus-specs/pull/2649 https://notes.ethereum.org/@djrtwo/ry_muOtEt From Leo (BSC) to Everyone: 02:48 PM : https://github.com/ChainSafe/eth2-crawler/pull/123 From parithosh to Everyone: 02:52 PM : Can someone buy a star and name it the same as some city? - merge both the naming schemes