2020-11-12 Eth2 Call 52.txt

#52 call - https://github.com/ethereum/eth2.0-pm/issues/191

Nimbus

Networking:
- Improve tracking of resources: streams, channels
- Gossipsub audit fixes (done late because 1.1 was not tested enough during its allocated phase)

Core:
- Toledo
- Documentation
- Infura HTTPS support (instead of just websocket)
- Faster ETH1 chain sync/monitoring
- Pre-release 0.6 “Hope” 2 days ago https://github.com/status-im/nimbus-eth2/releases/tag/v0.6.0 with prebuilt Linux binaries and a new release planned today.
- Gradually moving users away from makefiles and will provide precompiled executables
  for all major platforms (Windows, Linux, Mac and also ARM builds).

Release management:
- Preparing for mainnet
- Creating sanity checklists
- Creating mailing list
- Monitoring
- Support/coverage plan
- Lessons learned in directory renames (consistency) and permission changes (security)
  that were holding databases or were used in systems scripts
- Used to have a way to create deposits and validator keys from Nimbus
  To avoid confusion and fight scam attempts, this is now undocumented and developers-only. This way, user message can be “there is only one blessed way to deposit”.
  The Nimbus book will redirect to the official launchpad
  
——

Testing and release updates

BeaconFuzz:

found 2 consensus vuln on Prysm
- off by one in attestation processing
- wrong epoch when validating proposer slashings

Teku
- spec divergence on slashing similar to Prysm

Working with EF new team member for larger scale fuzzing

Custom fuzzing engine
for now mutation based buzzer is working
Blog post coming soon

Justin:
2 more people joining EF, still looking for security people at the EF
More RFP on testing coming up (networking, chaos testing, …)

——

Testnets: Toledo
including epoch with 100% attestations and participation

New testnet: Pyrmont
Need feedback from client if they can confirm Tuesday.
All OK for clients but couple days might be better.
==> Wednesday
-> mention on discord to not create too much validators.

Sunsetting Medalla, no one against

Vitalik: What testnet should we point people to?
-> Pyrmont. Will be 100k validators controlled by implementers/EF and then opened to public.

—> Have a way to update all users if something happens 1 hour before genesis

——
Client updates

Nimbus

Lodestar, Cayman
1.0 candidate, everything done besides Gossip 1.1

Prysm, Terence
Closed all issues with ToB audit, audit report is only
Improved on initial syncing process
Better at exploring forks during non-finality
Testing peer scoring
Slasher interchange format work
Issues to close before Nov 24th in milestones

Lighthouse, Paul
Blog post published about planned v1 release.

Teku, Anton
(…)
Fixed issues on OOM during long period of non-finalisation
Geth related fixes
Infura related fixes

——

Research updates

Aditya: Weak Subj Sync
Lighthouse Teku Nimbus support weak subjectivity sync from a block
Etherscan will support it
Beaconcha.in slightly after genesis

Vitalik: Data availability for phase 1

TXRX: Executable beacon chain proposal
so that Eth1 can work in the beacon chain

BSC: https://github.com/leobago/BSC-ETH2/blob/master/report/Clients%20behaviour%20report.pdf
- sync speed figures, including the number of peers connected
- Disk usage (with 35GB)
- CPU usage, note: 2 hyperthreaded cores
- Memory usage (5GB)
- Test done first week of Nov.

——

Networking updates

Adrian: Tested scoring parameters, works well, will merge to master

—— 

Spec discussion

———

Open discussion:

- Staking grant round
- Slashing interchange format
- Q: mitigation of “local time far in the future” to kick a validator via slashing protection
  -> no client team implement this yet.

———

Chat
From lightclient to Everyone: (3:48 PM)
re phase 2: is there an appetite to make a backwards incompatible change to the VM (to wasm or something else) in the next 3-5 years?

From Alex Stokes to Everyone: (3:50 PM)
one option on that front: deploy the new vm in a rollup and consider enshrining it as it demonstrates suitability