AllCoreDevs - Testing #082
Transcript
- Dima Gusakov | Lido
GM
- Mega | Lambda
gm
- Ivan | ethrex
gm
- Leny IHIRWE
gm
- Mario Vega
gm all 👋
- David (dionysuz)
g00d m0rning ethereans ☀️
- jochem-brouwer
gm
- danceratopz
gm!
- Potuz
GM
- Justin Traglia
Gm
- Jun Song
GM
- emma (reth)
gm
- FLCL
maybe rerun it again with just single CL + multiple ELs and add CL one by one
- terence
Ya I was gonna say that
- terence
Def
- Barnabas
Do other clients also downscore based on some colocation ip range thing?
- Parithosh Jayanthi
Do we have enough cpu limits to accommodate that?
- Barnabas
Replying to "Do we have enough cp..." Yeah we should be good
- FLCL
are those CL side issues?
- Parithosh Jayanthi
Could you guys tell us what would help you debug this better next time?
- Parithosh Jayanthi
What’s missing in the tooling?
- Parithosh Jayanthi
Forky updates?
- Nico Flaig
Replying to "are those CL side is..." EL devs might be interesting in checking why most of the ELs are optimistic
- Nico Flaig
Replying to "are those CL side is..." that stops us from attesting/producing blocks
- Justin Traglia
- FLCL
Reacted to "that stops us from a..." with 🤔
- Justin Traglia
- Tuyen
Replying to "Forky updates?" I’d like to be able to pull fork choice debug data from different nodes
- Ivan | ethrex
We started working on it last week but we saw no tests back then
- Barnabas
We shouldn’t increase the scope of glam-devnet-6 btw!
- Parithosh Jayanthi
Replying to "Forky updates?" So a forky that works? https://forky.glamsterdam-devnet-5.ethpandaops.io/
- Ivan | ethrex
Replying to "We started working o..." https://github.com/lambdaclass/ethrex/issues/6583
- Ivan | ethrex
Tracking issue ^
- Maria Silva
One sec
- Louis
Replying to "One sec" We can hear you now
- Tuyen
Replying to "Forky updates?" It’s much more convenient to see details of /eth/v2/debug/fork_choice from different nodes Right now I can only pull that data from lodestar nodes
- jochem-brouwer
@Mario Vega sorry did not look correctly, https://github.com/ethereum/execution-specs/releases/tag/tests-bal%40v7.2.0 has EIP-8037 tests
- Mario Vega
Reacted to "@Mario Vega sorry di..." with 👍🏼
- Marius Van Der Wijden (M)
If we have time at the end I would like to take 2 min for rest engine api update
- Barnabas
Reacted to "If we have time at the end I would like to take 2 min for rest engine api update" with 👀
- Ivan | ethrex
Reacted to "If we have time at t..." with 👀
- Justin Traglia
Replying to "If we have time at t..." Nice. Already in the agenda 🙂
- Edgar | ethrex
Reacted to "If we have time at..." with 👀
- Barnabas
Devnet numbers are imaginary
- Toni Wahrstätter
I'm not sure about the framing of "stable containers block higher gas limits". It's more, the constant caping deposits is blocking us, or, clients not being able to process that many deposits. I'd prefer bump the constant if the nr of deposits is the problem for scaling.
- jochem-brouwer
Antwoord verzenden naar "Devnet numbers are..." bal-devnet-7+5i
- Justin Traglia
Reacted to "bal-devnet-7+5i" with 😄
- Parithosh Jayanthi
Reacted to "bal-devnet-7+5i" with 😄
- Mario Vega
Reacted to "bal-devnet-7+5i" with 😄
- FLCL
does not max length affect merkleization? you want to omit it?
- Mega | Lambda
Reacted to "bal-devnet-7+5i" with 😄
- Louis
Reacted to "bal-devnet-7+5i" with 😄
- FLCL
Could you share a link to this pls?
- Marius Van Der Wijden (M)
Replying to "I'm not sure about t..." Agree that its the wrong framing, co-opting a different issue. The bigger problem imo is that it takes ages to verify them at the cl
- Etan (Nimbus)
Replying to "Could you share a li..." https://github.com/ethereum/consensus-specs/pull/4630
- Justin Traglia
Reacted to "https://github.com/e..." with 🙏
- Eitan Seri-Levi
Would stable containers make the devnets more stable
- FLCL
Replying to "Could you share a li..." Ty!
- Justin Traglia
Reacted to "Would stable contain..." with 😅
- terence
Replying to "Would stable contain..." lol
- FLCL
Reacted to "Would stable contain..." with 😅
- Dima Gusakov | Lido
Reacted to "Would stable contain..." with 😅
- Barnabas
Replying to "Forky updates?" Is this standard beacon api?
- Etan (Nimbus)
Replying to "Would stable contain..." it would make the tests closer to something that we eventually want 🙂
- Barnabas
Reacted to "Would stable containers make the devnets more stable" with 😅
- Bharath
Reacted to "Would stable contain..." with 😅
- Dima Gusakov | Lido
Apps devnets for sure
- jochem-brouwer
Heeft gereageerd op "Would stable conta..." met 😅
- Nico Flaig
Reacted to "Would stable contain..." with 😅
- Etan (Nimbus)
Replying to "I'm not sure about t..." the verification is separate from the serialization cap. even if the CL has a queue for the verification, we still need to lift the serialization cap
- NC
Reacted to Would stable contain... with "😅"
- Barnabas
More than 30 mins 😂
- Dima Gusakov | Lido
Reacted to "More than 30 mins 😂" with 🔪
- Justin Traglia
Reacted to "More than 30 mins 😂" with 🔪
- Parithosh Jayanthi
Reacted to "More than 30 mins 😂" with 🔪
- Edgar | ethrex
Reacted to "More than 30 mins ..." with 😂
- Edgar | ethrex
Reacted to "More than 30 mins ..." with 🔪
- Tuyen
Replying to "Forky updates?" V2 is under review https://github.com/ethereum/beacon-APIs/pull/615 v0 was in the spec for years
- Enrico Del Fante (tbenr)
Reacted to "More than 30 mins 😂" with 😂
- Eitan Seri-Levi
A week makes sense to me
- Mario Vega
Reacted to "More than 30 mins 😂" with 🥲
- Barnabas
Reacted to "A week makes sense to me" with 👍
- Toni Wahrstätter
Reacted to "More than 30 mins 😂" with 🔪
- terence
I personally will like to see builder api working but both can happen in parallel I guess
- Eitan Seri-Levi
Reacted to "More than 30 mins 😂" with 🥲
- Toni Wahrstätter
https://nerolation.github.io/hegota-eip-presentations/eip-7997.html
- jochem-brouwer
for call notes: Arachnid deterministic factory and Nick's method: https://github.com/Arachnid/deterministic-deployment-proxy/tree/master
- Pooja Ranjan
Reacted to "for call notes: Arac..." with 👍
- jochem-brouwer
Antwoord verzenden naar "for call notes: Ar..." Contract deployed on many chains at 0x4e59b44847b379578588920ca78fbf26c0b4956c
- Mario Vega
I love it when an option is basically “do nothing” and it’s also the best option
- Potuz
What would be an argument against option 2?
- Barnabas
Why was option 2 not even mentioned in eip 7997 ?
- jochem-brouwer
Heeft gereageerd op "I love it when an ..." met ➕
- Parithosh Jayanthi
Reacted to "I love it when an option is basically “do nothing” and it’s also the best option" with ➕
- Nico Flaig
Reacted to "More than 30 mins 😂" with 🔪
- frangio
Replying to "Why was option 2 n..." it was mentioned in the discussions
- felix (eest)
Reacted to "I love it when an ..." with ➕
- frangio
Replying to "Why was option 2 n..." https://ethereum-magicians.org/t/eip-7997-deterministic-factory-predeploy/24998/15?u=frangio
- frangio
Replying to "Why was option 2 n..." but the factory is also mentioned in the eip motivation
- Louis
Reacted to "I love it when an op..." with ➕
- Parithosh Jayanthi
Love it, adding stuff to genesis files is like a 1 line PR for us
- Barnabas
Replying to "Love it, adding stuf..." Its already done 😄
- Parithosh Jayanthi
Reacted to "Its already done 😄" with 😂
- Mario Vega
Reacted to "Love it, adding stuf..." with ❤️
- Parithosh Jayanthi
Replying to "Love it, adding stuf..." Even better
- Barnabas
Replying to "Love it, adding stuf..." https://github.com/ethpandaops/ethereum-genesis-generator/pull/294
- jochem-brouwer
arachnid has a hardcoded gas limit
- jochem-brouwer
Antwoord verzenden naar "arachnid has a har..." for the deployment at target address (it is fixed at 100k)
- jochem-brouwer
Antwoord verzenden naar "arachnid has a har..." the nasty thing is that the transaction takes account high gas prices, but not that the gas cost would be more than double (in gas units)
- raxhvl
Opstack supports this assumption laid by option 2 (https://github.com/ethereum-optimism/optimism/blob/dfea9e09dd72372f6c879d83eafc58b81caab0a2/packages/contracts-bedrock/specs/l2-upgrades-2-contracts.md?plain=1#L133)
- Łukasz Rozmej
EIP specyfies a super simple contract, why it is not deployable?
- Maria Silva
Aren’t the next topics EL?
- Barnabas
Replying to "Aren’t the next topi..." There would be still some CL topics to discussed regarding range sync
- Maria Silva
Reacted to "There would be still..." with 👍
- Barnabas
Replying to "Aren’t the next topi..." That’s pretty unrelated to ELs
- Barnabas
Replying to "Aren’t the next topi..." Just trying to maximize everyone’s time spent the best
- Toni Wahrstätter
Replying to "EIP specyfies a supe..." 7997 works, but the original nick's method tx has a tx.gaslimit of 100k and it doesn't work anymore to submit that tx
- jochem-brouwer
oh wow this is a good point
- Barnabas
Replying to "Aren’t the next topi..." CL devs focusing on CL issues, and EL devs focusing on EL issues
- jochem-brouwer
just use arachnid to deploy the "fixed" create2
- terence
Replying to "Aren’t the next topi..." Ya lets do that if possible
- Łukasz Rozmej
Replying to "EIP specyfies a supe..." But why?
- Barnabas
Yeah lets mod 7997
- jochem-brouwer
From process point of view I would say do this new EIP and preserve 7997
- jochem-brouwer
@justin you can skip my point
- Nico Flaig
https://github.com/ethereum/EIPs/pull/11760 ELs take a look
- Toni Wahrstätter
the state gas costs increased and now it runs oog
- Parithosh Jayanthi
Let me know when to create the breakout rooms 🙂
- Justin Traglia
Reacted to "@justin you can skip..." with 👍
- Łukasz Rozmej
ACDC - please update Ethereum (google) Calendar time!
- FLCL
Marius could you check comments
- Justin Traglia
Replying to "Let me know when to ..." Will do. Just one minute.
- Mario Vega
Reacted to "ACDC - please update..." with ➕
- Parithosh Jayanthi
Replying to "ACDC - please update..." Protocol r&D calendar?
- Justin Traglia
Reacted to "ACDC - please update..." with ➕
- Łukasz Rozmej
Replying to "ACDC - please update..." yes
- Tamaghna Chourdhuri
👀
- Parithosh Jayanthi
Replying to "ACDC - please update..." Ill see who has the permissions
- Łukasz Rozmej
Reacted to "Ill see who has the ..." with 🙏
- Jun Song
I’m looking into it (Prysm)
- Tamaghna Chourdhuri
Yeah prysm and nimbus are the targets
- Marius Van Der Wijden (M)
Reacted to I’m looking into it ... with "👍"
- Tamaghna Chourdhuri
Reacted to "I’m looking into it ..." with 👍
- Barnabas
Reacted to "Yeah prysm and nimbus are the targets" with 👍
- Nico Flaig
- David (dionysuz)
Replying to "ACDC - please update..." done
- Łukasz Rozmej
Replying to "ACDC - please update..." Ok it updated tfor this week, but not for future weeks?
- Tamaghna Chourdhuri
Replying to "I’m looking into it ..." Ill drop you a dm jun!
- Jun Song
Reacted to "Ill drop you a dm jun!" with ❤️
- David (dionysuz)
Reacted to "Ok it updated tfor t..." with 👍
- Łukasz Rozmej
Replying to "ACDC - please update..." Is this by design?
- David (dionysuz)
Replying to "ACDC - please update..." yes
- Parithosh Jayanthi
Replying to "ACDC - please update..." Yes its a one time thing, we are trying out the new time
- Parithosh Jayanthi
Replying to "ACDC - please update..." Not committing to it
- Potuz
Do we have a consensus one?
- Potuz
How do people find the places
- Parithosh Jayanthi
Ill sign you
- Parithosh Jayanthi
Ill asign you
Call summary
Targets
- •Wednesday — Glamsterdam devnet-6 relaunch (same spec as devnet-5) - 00:13:50
Decisions
- •Glamsterdam devnet-5 abandoned; devnet-6 relaunch on Wednesday with same spec - 00:13:50
- •EIP-7688 (stable containers) deferred until EPBS devnet stable for at least a weekend; targeting devnet-8 - 00:26:41
- •EIP-7997: Option 2 adopted — enshrine Arachnid at known address in Genesis; EIP or 7997 modification to follow - 00:38:20
- •Breakout rooms split: EL on deployment/8037 topics, CL on range sync and EPBS issues - 00:40:52
Highlights
- Organizational:
- ·ACDC this week at 11:00 UTC (3 hours earlier than normal); one-time change, not committed - 00:41:31
- ·SSZ Engine API: Geth and Erigon implementing; EF intern targeting Prysm and Nimbus on CL side - 00:41:43
- Testing Progress:
- ·Glamsterdam devnet-5 struggling badly; multiple clients forking for unclear reasons - 00:06:09
- ·Prysm unable to sync from finalized checkpoint; isolated and out of peers due to Hetzner co-location downscoring - 00:07:31
- ·Potuz: no value in recovering devnet-5 without understanding root cause; relaunch preferred - 00:08:02
- ·Most ELs optimistic on devnet-5; cannot produce blocks or attest — recovery unlikely - 00:11:49
- ·bal-devnet-7 running stable; benchmarking underway, numbers looking good per Stefan - 00:05:17
- Fork Status And Schedule:
- ·Glamsterdam devnet-6 relaunch targeted for Wednesday; same spec as devnet-5 - 00:13:50
- ·Glamsterdam devnet-6 scope unchanged; new EIPs targeting devnet-7 - 00:21:31
- ·New consensus-specs Alpha 10 and Beacon APIs V5 Alpha 2 released; target for devnet-5/6 - 00:22:15
- Eip Proposals Glamsterdam:
- ·EIP-7688 (stable containers): blocks 200 MW tests due to SSZ deposit limit overflow; only Nimbus and Lodestar fully implemented - 00:23:08
- ·Consensus: defer EIP-7688 until EPBS devnet stable for at least a weekend; target devnet-8 - 00:26:41
- ·EIP-7997 (deterministic factory): Option 2 preferred — enshrine Arachnid contract in Genesis file; no-op for mainnet - 00:30:22
- ·Arachnid cannot be deployed post-Glamsterdam via Nick's method; 8037 increased gas costs past 100K tx limit - 00:36:06
Action Items
- •Toni Wahrstätter / Mario Vega / frangio - Toni to align with Mario and frangio async on EIP-7997 vs. new EIP approach for enshrining Arachnid - 00:39:32
- •CL teams - Additional CL client (beyond Prysm/Nimbus) to volunteer for SSZ Engine API implementation - 00:43:00
Key decisions
Glamsterdam devnet-5 abandoned; devnet-6 relaunch on Wednesday with same spec
SFIdevnet-5 was struggling badly, multiple clients forking for unclear reasonsEIP-7688 (stable containers) deferred until EPBS devnet stable for at least a weekend; targeting devnet-8
Unknownblocks 200 MW tests due to SSZ deposit limit overflow; only Nimbus and Lodestar fully implementedEIP-7997: Option 2 adopted — enshrine Arachnid at known address in Genesis; EIP or 7997 modification to follow
PFIOption 2 preferred — enshrine Arachnid contract in Genesis file; no-op for mainnet