Dencun Interop Testing #30

Transcript

foreign [Music]

thanks Puja hey guys uh Tim is off today so unfortunately I'm the host of this call Welcome to the Dan interop testing call number 30. hopefully tub will be back next week for you Barnabas agenda looks pretty light but I would like to hear what's going on with devnet8 does anybody want to share an update I noticed there's been some chatter in the interop channel uh yeah I can give a quick update so this morning we saw some basil issues and uh ethereum JS issues so both of these clients and two had some issues

thinking up to head they do uh had some trouble with The Blob hashes and transaction hashes where it could not propose a block because they had some mismatched hash maybe someone from basican give an update I'm not sure is anyone here from but it doesn't seem yeah it doesn't seem like it no no but they they pushed out the update and attempt to have fixed the issue so now they are up to head and also they seem to be able to propose again great and the East JS issue is that still being investigated

this issue was mostly related to sync and uh basically uh generally what really happens is that each JS sync is not really production quality and we are working on it but sometimes it gets stuck and one needs to sort of uh restart the client or resync the client and that gets its result that was the main problem but uh there the other problem that I noticed was with regard to Fort mon because Fort moon for example at certain blocks shows that JS has a different block hash than the other clients and when I sort of did

uh get blocked by a number uh get blocked by number only JS as well as other clients so that hash was matching so there is also for there there is another issue that we discovered in Fort mon with Barnabas is that the hashes that Pokemon is showing does not correspond to the actual blockage that is over there so there is a variance in what forkmon is showing as well as what the actual block hashes are as well as the second issue with Fort mon is that for the same block for EGS uh if you do get blocked by a number then the hash is correct and is matching

with what other clients do get locked by number but somehow I'm not sure how for Pokemon is collecting that data and where the issues okay got it um yeah as you reported these issues to forkmon okay okay maybe yeah maybe this week it's not really urgent because uh in general it still shows that there's a mismatch but uh just showing the wrong uh hash yep got it

cool uh how did the devnet look last week I noticed there was uh yeah somebody had posted an image of the beacon chain website and there was a period where we were missing a lot of blocks is that the same issue that has sort of been resolved or that was a different thing um with the tennis shoes I don't think we had issues last week Oh I thought that sorry I thought this someone had posted an image of the devnet aid Beacon chain missing like 50 blocks for a period of time no it was a specific line it wasn't like okay it was one specific bag that was not uh

yeah I think it was busy so the busy fix is uh you know yeah sort of result Lighthouse still has some issues proposing blocks but all right we'll see makes sense I'm trying to come on uh into the call from Brazil right now okay let's Circle back to that if somebody pops into the call um there's another issue with pool transaction message ROP encoding is this similar to what we discussed a while back with the East 68 release and changing for the future transaction

um do you know which clients are sending which ones is it the list versus a wrapped issue maybe we have different encodings uh Netherlands and the person and I saw a similar issue in practice forget so all three clients have different encoding here maybe two sure

are you just referring to the test that's failing on hive Cancun or are you talking about something you're seeing specifically because I know the test that's failing for us on hive Cancun is due to 4kid mismatch which yeah that was uh that's something that there's a PR to fix right now I don't think it has to do with the ROP encoding oh I'll let you find them um and this whole bit from your mind received from uh zubox okay fantastic uh I think Barnabas or Mario your audio

is really low oh it looks like Barrio yeah I can't hear you very well good morning Terence good morning sorry I'm late I almost forgot there's a meeting today it's a holiday yeah you're forgiven um okay so there is the ROP was not part of the last PR oud cool

um yeah let's yeah just post on interop if there's something specific we want to continue discussing about that was there any other devnet 8 related things that people wanted to discuss um could we merge 398 [Applause] it's been open for quite a while it would be nice to merge that yeah yeah I'll look into that today um is there a timeline for time uh for

that nine that clients would like to yeah that's that's what I was also kind of curious about uh I mean just for like guests perspective I feel I don't know maybe yeah next week we could have everything ready for devnet9 but yeah Merrick yeah so we want to test syncing with all different client players not only Netherlands and it is currently running so expect that we will share some results but before we finish those testing maybe a little better to not stop does not egg I'm not sure what what is it should be anytime

soon yeah got it uh what is the ETA of completion sometime this week yeah I I think so uh basically we are running all client first right now and we are waiting for uh got it um yeah I I think like maybe shooting for September 12 seems like a reasonable goal yeah what do people think I know Terence you were questioning the timeline that Dan Craig was talking

about on all core devs I don't know what you think about devnet9 next week yeah my for local devnet I don't really have much input I think yeah next week is fine I'm just yeah my feedback was more on like overall timeline whether it's that 2023 or already 2024 but yeah we don't have to get into that this meeting but I guess I want to confirm only consensus players died but that day and night I don't think there's any change can anyone like confirm this with me that is there any questions to their client team I just want to make sure I'm not running here yeah that's my understanding also it

seems like the big difference between a to nine is that we'll not enshrine 4788 system contract yeah then from us there's yeah I mean we just have one or two bot faces we can easily get in uh by the end of the week so yeah um next week sounds good to me it would I I know it's quite unrelated but it would also be called uh to have the holoca releases by that the latest from each client teams because uh we will want to make sure that husky is ready to go by the 15th right maybe even sooner would be

probably better it's quite unrelated nine yeah so definitely need to have the whole ski releases out probably by the end of this week yeah preferably how about we post on Discord uh saying that we'll try and Target September 12 for a devnet 9 launch and see if there's any feedback on that date and if not we just try and go for that yeah I think that's good and I think we should push for September 7th for client releases sounds good I will sounds good I'll post

a message here right after this call and get some feedback any preparances for well-defer sizes by the way because on dev Note 8 we run a very very low number of validators just so it's easy to add new ones and remove ones but I'm just curious if you want to do a bigger test or if you're just going to leave the bigger test for all this key Pulaski did you say a Pulaski not up checking if checking if everything works just kidding um yeah I don't have a preference

I don't have any preferences anyone have a preference on this I do have a preference on like we use some sort of Builder and relayer for definite night before using a test net just because I feel like everything that I think will break I don't I'd rather break on devnet without doing tests and answers the question is uh the Builder and the relay Readiness I'm not so sure about that and whether even I think husky is the first test net and whether there will be Builder and relayer at Host key uh we are currently doing uh Builder testing on kurtosis so hopefully it's going to be ready by next week I can I

cannot promise it uh what we're gonna be trying to push for is a capella starting for that nine so we're gonna be starting uh Shanghai and Genesis time is going to be the same and this is also not something that the med software supports right now so this is going to be a breaking change for them also but yeah I totally agree that we would want to onboard Mev before we Fork any distance we need to make sure that maybe it works before we start messing around with Husky Mario yeah so one thing that I noticed is that

I was testing the mock Envy Mark builder in the grotosis setup and the problem there was that yeah the Meb boost that sits in between the CL client and the mug Builder and the Builder I was writing um it was not ready for dinner at least checked last week and what was happening was that my Builder was building the nap um payloads for the CL and then when it send them to the med boost and then the BMV was relate to the cl to the CL client it's stripped out all the the nav related uh Fields so basically uh what was uh ending up in the CL was just the

uh the Shanghai payload instead of they can go one um so that's a problem I don't think we can I mean unless we remove any V Boost from kurtosis and from the devnets which I don't think it's really good idea anyway to properly test we cannot test unless any Vehicles is updated uh yeah yeah I mean I assume we want to get Mev boost updated up like that's that's the whole point that they need to support the map otherwise yeah we are kind of altered uh who's the main

people that we need to talk to about getting them to support and then now like this seems like it needs to be kind of their number one priority is it just metacris or is there anybody else that would be working on this we do have a channel for flashbot and let me reboost and uh people just quite unresponsive so we've been trying to push for this in the past two months and yeah we haven't gone we haven't got as much support as we were hoping for I see but it's mostly just like the ball is in their court at this point and we're asking them okay exactly

uh let me make a note to follow up again with them all right yeah I guess that was kind of the last item also on the agenda was just talking about Builder testing readiness it seems like we're keep doing some testing in kurtosis I mean I guess I don't really understand like if and that boost doesn't support the NAB then what exactly are we testing with kurtosis well the flashbot guys are using kurtosis to test

the MEP boost it's tough so they are the ones that are pushing for this so they can easily spin up a new locally on their machine and and see if then if they can make Mev compatible with the map okay so the NAB I see so the NAB work is ongoing for them and they're testing okay I see so there's an open PR uh regarding this I think this one add the NAB types and Flow

wait I yeah it's for quota system once you manage to get it working in criticize it then uh you can be quite confident that it's going to be supported also for definitely got it okay uh let's follow up with it again with them try and add a little bit of pressure and yeah they need to be on devnet9 I don't think I agree we need to not uh move to a public testout before we're running the Builder on the devdet yeah so another update is that we are currently working on supporting the mock

builder at least on high so we can have regression testing there too um basically just uh sort of Sanity testing where we um verify all the signatures that the cl sends back to the Builder and all the all of that stuff also including and those kind of Sanity desks and Sanity checks as as in the form of regression tests in High um so there's that too we we can have that in parallel to kurtosis and in part two to the night makes sense this is testing specifically the Builder API of the CLS right yeah

yeah so so the approach has been to have um like the only the the Builder API interface tested in High um I don't know yeah if anyone wants to um chime in and let us know if we should include any new Boost but I think this is another approach of the testing without I mean it was just to test the interface and build their API the the errors that can happen from the Meb from the from the Builder also um that's that sort of the mentality that we are going uh with hype testing for the Builder right now yeah we've talked about this a couple times and I I do think this is the right

approach for Hive I don't think should be doing integration testing on every piece of the Mev stack and hive um so yeah if we have that input that implemented in Hive and then we're able to run some of this builder of boost software on the dev nuts and maybe in kurtosis that's yeah that seems like the right amount of coverage that we should be providing for that software yeah and another thing is that it's um it's a little bit more uh robust uh right now the map Builder as compared when it was testing Shanghai we are now verifying every single signature um this is related to the issue that was

less found in the uh with me boost and the and and one of the sealed Banks so right now we are trying to uh enforce checking every single thing that we can on the on the mug Builder to to make sure that nothing slips great okay that's most of the agenda that was set on the issue is there any other topics that people were hoping to discuss today maybe just to point out um some existing open PR's um for Hive

one of them is the interrupt Builder we're working on and the other one is just to improve the execution client engine API coverage some of the clients have pointed out some coverage that is missing so we are including more tests in in these two PRS that I submitted in the in the in the BM issue also we're working on python um dispenser and pointed out some of the PRS you're working on I don't know if you want to uh just mention uh yes so from the execution spec test sites and Dan's been working on the transient

storage tests for EIP 1158 and the should hopefully adds some more coverage in addition to the tests that I believe Dimitri added in the original theorem test repo and we've also been working on updating the framework to generate invalid field tests for the engineer at payload Fields So currently when we generate tests we don't have a wave generating them with fields that are deemed invalid by transaction tools and for example another phone unit 64 fields and so we've been working on generating these types of tests specifically for the Duncan fields and then ideally wants me that he's working and we'll run these in the uh

simulator I think uh Mario also added a test for and they can read deployment and so this will be it for definite nine essentially tests that the expected leaping group contracts if not president before the camping Fork um and checks that the storage is updated and then yeah just to kind of I guess orchestrate with um when everyone wants to um change the definite name and we have tests ready for it so if September 7th is good tests and it's Thursday

awesome thanks for those updates guys I have a question with regard to how the 4788 contract will be deployed I we added 3ngs are not clear about it because the sender that is there we have the credentials for it or will we deploy it using some sort of a system here in the EAP there is actually a complete transaction body there so you can see that the signature is included in that Json so if you're able to read the Json value and encode it

into ROP it can actually just be sent over the network and it will be included in a block once the sender account is funded with eth so it's not a transaction that we have the private key or it's not from an account that we have a private key for it's more that the signature was chosen and then Italy the using the EC recovery able to figure out who the sender would be so it's kind of a one times yeah signature to send this this transaction well any last comments or questions on

dencoon okay let's call it I will post this message on all Court devs just mentioning the request for a September 7 client I think it was September yeah September 7 client release for holeski sorry September 8th uh date for client releases for holeski and targeting a September 12 launch of devnet9 and we'll see if there's any feedback on those dates thanks a lot for joining we'll talk to you all next week

bye guys cheers