AllCoreDevs - Testing #093
Transcript
- Pooja Ranjan
gm
- Justin Traglia
- Barnabas
- Barnabas
head votes have been pretty bad even earlier.
- Barnabas
not as bad as lately tho
- Stefan Starflinger
- Justin Traglia
I did some bid spamming tests last week. Around 300-500 bids per slot for a few hours.
- Barnabas
do we have dedicated hive tests for eth estimateGas?
- Barnabas
I feel like estimateGas is gonna be very valuable post glammy
- Chase Wright
Yes but not for Amsterdam, hivechain needs to be updated to support Amsterdam
- Parithosh Jayanthi
Is main branch support the blocker there?
- Chase Wright
No, it’s getting people to review hive PRs lol
- Barnabas
Replying to "No, it’s getting peo..." who’s blessing do we need?
- Toni Wahrstätter
Will we do a devnet-9?
- Parithosh Jayanthi
Replying to "No, it’s getting peo..." felix
- Chase Wright
Replying to "No, it’s getting peo..." Historically, Felix
- Stefan Starflinger
Reacted to "Will we do a devnet-..." with 👍
- Chris Berry
Reacted to "Will we do a devnet-9?" with 👍
- Luca | Vero
Reacted to "Will we do a devnet-9?" with 👍
- Mario Vega
Reacted to "Will we do a devnet-9?" with 👍
- Chris Berry
We agreed on Thursday there would be another devnet
- Barnabas
devnet 9 can be non finality imo
- Nico Flaig
Reacted to "devnet 9 can be non ..." with 👍
- Justin Traglia
Reacted to "devnet 9 can be non finality imo" with 👍
- Barnabas
Replying to "We agreed on Thursda..." is it ok if its not long lived?
- Stefan Starflinger
Reacted to "devnet 9 can be non ..." with 👍
- cayman
Reacted to "devnet 9 can be non ..." with 👍
- Toni Wahrstätter
then the question is, do we need a devnet-10?
- Nico Flaig
Reacted to "then the question is..." with 😂
- Parithosh Jayanthi
Reacted to "then the question is, do we need a devnet-10?" with 😂
- Stefan Starflinger
Reacted to "then the question is..." with 😂
- Justin Traglia
Replying to "then the question is..." I hope not lol
- Barnabas
Replying to "then the question is..." devnetflation
- Stefan Starflinger
Replying to "then the question is..." we start back at 01 after 09
- Barnabas
Can we aim to to have trunk support by ACDE this week from every single client?
- Stefan Starflinger
Reacted to "Can we aim to to hav..." with 🛳️
- Justin Traglia
Replying to "then the question is..." Let’s do base 16, devnet-A 😄
- Stefan Starflinger
Reacted to "Let’s do base 16, de..." with 🤔
- Justin Traglia
Replying to "then the question is..." Kidding of course
- Barnabas
Reacted to "Let’s do base 16, devnet-A 😄" with 🤔
- Parithosh Jayanthi
Reacted to "Can we aim to to have trunk support by ACDE this week from every single client?" with 🛳️
- Toni Wahrstätter
Reacted to "Let’s do base 16, devnet-A 😄" with 🤔
- Justin Traglia
Link to the CL breakout: https://ethereumfoundation.zoom.us/j/89441658268?pwd=yjqf4TheEb1R7ZSuesCdOAy5tdyVk7.1
- cayman
Reacted to "Can we aim to to hav..." with 👍
- FLCL
Reacted to "Can we aim to to have trunk support by ACDE this week from every single client?" with 🤔
- Mario Vega
- Karim T. (matkt)
Do we have hive tests that validate this one ?
- Barnabas
lets aim to merge/close in all open prs today pls
- Chase Wright
Reacted to "lets aim to merge in all open prs today pls" with 😂
- Chase Wright
Merge all my slop hehe
- Stefan Starflinger
Reacted to "Merge all my slop he..." with 🛳️
- lightclient
which PR needs to be merged?
- lightclient
for hive chain
- Parithosh Jayanthi
Reacted to "Merge all my slop hehe" with 🛳️
- Stefan Starflinger
can we get the hive chain pr merged or get a fork so we can get testing started?
- spencer
this is hivechain: https://github.com/ethereum/hive/tree/master/cmd/hivechain ?
- lightclient
Reacted to "this is hivechain: https://github.com/ethereum/hive/tree/master/cmd/hivechain ?" with 👍
- Mario Vega
Reacted to "this is hivechain: https://github.com/ethereum/hive/tree/master/cmd/hivechain ?" with 👍
- Chase Wright
- Barnabas
do we know when felix is back?
- Chase Wright
- lightclient
i can look into it
- spencer
Reacted to "i can look into it" with ❤️
- Toni Wahrstätter
Reacted to "i can look into it" with ❤️
- Mario Vega
Reacted to "i can look into it" with ❤️
- Toni Wahrstätter
Replying to "i can look into it" *insert chad meme*
- Stefan Starflinger
Reacted to "i can look into it" with ❤️
- Mario Vega
- lightclient
Reacted to "*insert chad meme*" with ❤️
- Barnabas
can’t we just add branch support for hive?
- Chase Wright
Who? Felix is OOO lol
- Mario Vega
- Barnabas
Replying to "Who? Felix is OOO lo..." just ping EF devops and they will create whatever branches you need.
- Mario Vega
- Chase Wright
- Barnabas
Replying to "https://github.com/e..." @lightclient can you review?
- spencer
personally I like the blocking before all tests pass on hivechain, we should try to improve the hivechain update process
- Chase Wright
Reacted to "personally I like the blocking before all tests pass on hivechain, we should try to improve the hivechain update process" with ❤️
- Barnabas
can we rip out the go-ethereum dependency from hive lol?
- spencer
there is also the question of removing geth as the dependency of hivechain
- lightclient
Reacted to "can we rip out the go-ethereum dependency from hive lol?" with 😆
- lightclient
Reacted to "personally I like the blocking before all tests pass on hivechain, we should try to improve the hivechain update process" with ❤️
- Csaba
Which one is the geth PR in question?
- Chase Wright
- Csaba
I’ve joined late
- Barnabas
Replying to "Which one is the get..." https://github.com/ethereum/go-ethereum/pull/35397
- Mario Vega
https://github.com/ethereum/pm/issues/2195#issuecomment-5395878309
- jochem-brouwer
These bugs caused by parallelization are rather scary
- jochem-brouwer
Do clients if they use parallelization but get invalid block, attempt to re-execute it sequentially? To avoid race conditions in the parallelization
- spencer
Reacted to "Do clients if they..." with 💯
- Toni Wahrstätter
Replying to "Do clients if they use parallelization but get invalid block, attempt to re-execute it sequentially? To avoid race conditions in the parallelization" there's no race conditions in parallelization. All transaction can execute in a vaccuum. What is invalid when executing in parallel MUST be invalid when executing sequential
- Mario Vega
EELS has basically no RPC endpoint, we could create one though, but it’s not going to happen overnight
- Ben Adams
Reacted to "there's no race conditions in parallelization. All transaction can execute in a vaccuum. What is invalid when executing in parallel MUST be invalid when executing sequential" with 👍
- Guru
Reacted to "EELS has basically n..." with 👍
- spencer
Reacted to "EELS has basically..." with 👍
- Barnabas
Reacted to "EELS has basically no RPC endpoint, we could create one though, but it’s not going to happen overnight" with 👍
- Stefan Starflinger
Reacted to "EELS has basically n..." with 👍
- Csaba
Replying to "Do clients if they u..." I agree, if you get an invalid block, you would get it too with serial. The two match, or you have an implementation error
- Ameziane Hamlat
Reacted to "there's no race conditions in parallelization. All transaction can execute in a vaccuum. What is invalid when executing in parallel MUST be invalid when executing sequential" with 👍
- Csaba
Replying to "Do clients if they u..." Moreover, the introduction of helpers for parallel execution (BAL) is motivation to increase the gas limit. Serial re-execution would break that logic anyway.
Call summary
Targets
- •Thursday (ACDE week) - all CL client trunk branches merged; devnet-9 launch targeting early following week - 00:14:05
Decisions
- •Thursday deadline for all CL clients to have trunk branches ready; devnet-9 targets launch early next week - 00:14:50
- •devnet-9 will serve as the non-finality network, with fork transition tested before chaos begins - 00:11:57
- •HiveChain GoEthereum dependency will not be replaced before Glamsterdam; EELS RPC support to be prioritized for future forks - 00:41:04
Highlights
- Organizational:
- ·Pooja requests EIP authors approve PR #11855 to move Glamsterdam Meta EIP to review status - 00:10:58
- Devnet Planning:
- ·devnet-9 proposed as the non-finality chaos network; fork transition testing to be batched in before chaos phase - 00:11:57
- ·Teku and Lodestar trunk branches ready; Nimbus likely ready; Prysm and Lighthouse still pending - 00:13:46
- ·Deadline set: all clients must have trunk branches merged by Thursday (ACDE) to target devnet-9 launch early next week - 00:14:05
- Testing Progress:
- ·Glamsterdam devnet-8 near 100% participation; head vote correctness degraded to 65–80%, suspected Grandine attestation timing issue - 00:05:17
- ·Toni's malicious-strategy tests caused a Nethermind/Prysm fork and a Besu/Teku missed slot; posted to interrupt channel - 00:06:47
- ·Stefan's RPC interoperability report for devnet-8 highlights inconsistencies in eth_estimateGas; clients asked to review - 00:08:03
- ·Pawel's fuzzing of Glamsterdam spec surfaced bugs in blob access list and EL gas repricings; some fixes merged, remaining issues are partial blockers to continued fuzzing - 00:34:24
- El Breakout Hivechain Blocker:
- ·HiveChain (GoEthereum-backed test chain generator) not updated for Glamsterdam; blocks writing and merging all RPC compat tests - 00:27:09
- ·eth_createAccessList blocked separately: GoEthereum uses a different default gas value than all other clients; fix PR open against GoEthereum (#35397) - 00:29:45
- ·Discussion on removing GoEthereum dependency from HiveChain; EELS RPC endpoint proposed as long-term alternative but not feasible before this fork - 00:32:39
Action Items
- •lightclient - Review and unblock HiveChain Glamsterdam PRs (#1589, #1585) so RPC compat tests can be written - 00:21:42
- •spencer / execution-specs team - Cut new execution-specs release containing Pawel's fuzzing bug fixes (EIP-37 bug + 2610 removal) - 00:37:29
Key decisions
Thursday deadline for all CL clients to have trunk branches ready; devnet-9 targets launch early next week
devnet-9Deadline for CL client readiness to target devnet-9 launchdevnet-9 will serve as the non-finality network, with fork transition tested before chaos begins
devnet-9Definition of devnet-9's purpose as a chaos networkHiveChain GoEthereum dependency will not be replaced before Glamsterdam; EELS RPC support to be prioritized for future forks
Decision on HiveChain architecture for the upcoming Glamsterdam fork
AI Disclaimer: Some content or metadata on EIPsInsight may be AI-inferred or automatically compiled. If you find any discrepancy, please contact us at dev@avarch.org.