Eth2 Implementers’ Call #49 - 2020-10-01

[Quick contemporaneous notes by Ben Edgington]

Agenda: https://github.com/ethereum/eth2.0-pm/issues/184
Livestream: https://youtu.be/IRWQUQfq7yQ

Testing and Release updates

Danny is working on a v1.0 candidate release. There is a WIP pull request.

Protolambda has been working on a framework using yaml files to set up test networks with different clients, but stalled due to higher priorities.

Mehdi just published a new blog post on the differential fuzzing work. Found two critical bugs in the Blst library, now patched. Only targetting the Rust bindings currently. Found two minor discrepencies between clients around slashing. Not exploitable, and seeing fewer exploitable vulnerabilities in general. Teku is now added to the fuzzing framework. Starting work on custom fuzzing engine: faster, more flexible. [Jacek] Which BLS implementations will you test? A: Focusing on Blst as the majority used. [Mamy] Note that Lodestar is using MCL. A: hope to get to this eventually.

Spadina retro

Spadina now looking stable, but had only 1/3 participation at genesis due to various issues during the process. We will be doing another run: Launchpad up today, deposit contract early next week, genesis a week later (Zinken).

Afri suggested a call around release management of clients to discuss best practices. Target early next week. Danny and Afri to coordinate.

[Afri] Two days genesis delay seemed too short. [Danny] Agreed. Zinken network will have a 4 day delay. Proposing a 7 day genesis delay for mainnet. [Mamy] We need to communicate the delay better to users. Danny is putting out a blog post around Zinken that should help explain things. [Dankrad] Client teams need effective communication channels with users around the genesis time in case quick updates are required. Email notifications are effective. Could display comms channels on client start up.

[Hsiao-Wei] has collected some feedback notes on Spadina. Ethstaker Discord has been very helpful prompting a checklist around the launch process. Will share next week.

Client updates

Prysm
Did a Spadina postmortem and published it. Need better prep, docs, comms. Created a readiness checklist.

Up to date with v0.12.3. Added weak subjectivity checkpoint. Voluntary exit is now in place. Getting feedback from the latest audit and improving testing.

Nimbus
Last phase of audit is done. Fixes and fix review next. Significantly improved stability. Ethstats.io shows Nimbus uses least memory. Polishing user experience, logging, diskspace. Weak subjectivity sync is in PR.

Lodestar
Tried to participate in Spadina with 4 nodes. Node is still not stable: created genesis, but couldn't keep up with head.

Added WS chainstart: provide state and start from there. Finished standard API. Lots of refactoring and cleaning. Gossipsub v1.1 almost there. Slashing protection interchange and disc v5.1 are next.

Lighthouse
Spadina launch was smooth for the most part. Working towards pre-mainnet audits next week. New release coming with standard API. Slashing interchange format. Remote signer. Web UI. Weak subjectivity checkpoints.

Up to date with disc v5.1, but not yet integrated. Testing gossipsub v1.1 parameters. UPnP support done.

Teku
The main things we've been working on have been

• Discovery v5.1 is basically done, just catching up with the last few updates to the discv5.1 spec.
• A lot of work on weak subjectivity sync. We now have hidden CLI options to handle the checkpoint. We also have internal weak subjectivity checks in place. This is, checking the sync against the given checkpoint, checking that our peers are consistent with the checkpoint, and handling the case that the last finalised state is outside the weak subjectivity period. Next step is to be able to start syncing from a given checkpoint.
• We've been implementing some validator performance metrics
• And of course, preparing for and participating in the Spadina launch.
• We're also digesting the draft audit report that we got yesterday. Nothing too scary there.
• Next is standard API and validator split out.

Research Updates

On Ph1 spec, there's still a PR that's waiting until after the v1.0 spec release. There are now Ph1 tests. [Mamy] Note that these are 3GB big when uncompressed and cause problems with disk space: may be better to split out Ph0 from Ph1. [Proto] Immediate solution could be to compress the states: they contain a lot of zeroes.

[Leo] Continuing work on a network crawler based on Rumor. Making progress. Will post a link. Have found small bugs and proposed some features.

[Danny] On Eth1/Eth2 merge work: have identified a list of 5-6 items for further research.

Networking

Felix is trying to get the v5.1 discovery spec finished asap.

Spec discussion

Nothing.

Testnets

Danny will put out a blog post about Zinken today.

---

Chat highlights

From danny to Everyone: 03:03 PM

https://github.com/ethereum/eth2.0-pm/issues/184

From Mehdi Zerouali to Everyone: 03:10 PM

https://blog.sigmaprime.io/beacon-fuzz-08.html

From Hsiao-Wei Wang to Everyone: 03:15 PM

https://notes.ethereum.org/w4E0FB4nSzWPRj9pEeguhQ

From Mamy to Everyone: 03:16 PM

Is it fine to post it on https://github.com/eth2-clients/user-experience?

From Hsiao-Wei Wang to Everyone: 03:17 PM

re: Mamy sounds good!

From Leo BSC to Everyone: 03:31 PM

https://github.com/leobago/BSC-ETH2/tree/master/rumor/medalla-crawler