All Core Devs - Consensus #156

Transcript

Alex Stokes: Let's see, Josh, are you helping with the stream?

Josh Davis: Yep, whenever you're ready.

Alex Stokes: Okay, yep, let's do it.

Josh Davis: Alright! You're live!

Alex Stokes: Okay, everyone. This is Acdc, number 1, 56. It is issue 1490 in the ethereum. Pm Repo. I'll paste a link there in the chat. And yeah, today's call might be pretty short. I think the main topic on everyone's mind at the moment is Petra. So yeah, let's just go ahead and dive into Petra then. So the 1st thing we did have a mainnet shadow fork. Does anyone have any updates yet? I'm not sure that was, did you want to share your screen? Okay, anyway? Anyone have any updates on the main shadow Fork? Perry couldn't make it. But I did have an Async update. And he said, Yeah, the update is that the main shadow fork went mostly fine. There were some deposit problems on some clients, but it's due to the Shadow Fork config and can't happen on Mainnet.

Also, another nice thing is that notices forked picture yesterday. It seems like everything went well there any? Oh, yeah. Barnabas.

Barnabas Busa: Yeah, so sorry. I tried to turn on my mic before. So basically, yeah, diminish. For when kind of as expected we use the same deposit contract address as we use for may not. And this is what caused all the issues. So in previous shadow forks. This has not happened because we were using custom deposit contract address. And that's why it didn't happen in previous shadow forks. So.

Alex Stokes: Gotcha.

Barnabas Busa: Basically basically what what's wrong with the previous folks in, for example, in Holski is what made this one break. Basically. So the whole deposit queue had to be reprocessed and invited by every consensus layer client and the different different client. Different clients acted differently. Basically.

Alex Stokes: Gotcha. Okay? Well, it sounds like, this is kind of a known issue, and we can still proceed as planned. So then, speaking of. I was just trying to get the announcement blog post. So yeah, I assume at this point, everyone has the right client version there. It's a bit late in the game if you don't. But yeah, I guess. Just please raise a hand if that needs to change. Otherwise, yeah. One final reminder or our instant response plan here. And I think at this point we have info for everyone, except maybe Grandina. So yes. Are you interested in joining this.

Sorry I missed easier here.

Saulius Grigaitis: Yeah, yeah.

Alex Stokes: Yeah, we were. We were just having point people per client. In the event, something goes wrong with the fork and say, there's an issue with quarantine.

Saulius Grigaitis: Oh, yeah. Yeah.

Alex Stokes: Touch, with.

Saulius Grigaitis: Yeah, no problem. I'll I'll put the persons in the list.

Alex Stokes: Okay. Great. Thank you.

Saulius Grigaitis: Sure.

Sean Anderson: So on lighthouse we released 7.0 point 1 yesterday, which? It helps in some cases with state cash misses. So it's I. I think we'd have a slight preference for using that on the release page. But it's not like hugely mandatory. It affects some users.

Alex Stokes: Gotcha. Okay, that update should be pretty easy to make. So yeah, I'll make sure that happens. Okay, Terrence, would you like, is this 6 0 1 on top of 6 0.

Terence Tsao: Yeah. So we'll release 6.0 dot one today, which has 2 primary but fixes that helps in terms of like reorg. So previously. If a beacon, node, reorg, or like across ep boundary, the duty changes, but then the valid client will not be able to adjust to this change. So we fix that in the latest release after thorough testing, and then we fix a a pillow attribute event streaming endpoint, issue over the weekend as well. This is to help the re there, and the builder monitoring the event feed. So we face 2 bugs, and those will be on top of 6 0 dot one. It's not mandatory. It's a nice to have, but we just want to put this out there. People want to upgrade can upgrade, and those will be released today.

Alex Stokes: Okay, great. Yeah. Sounds like we might want to hold off on changing the blog unless it's critical. Yeah. So either way, so one thing.

Tim Beiko: One thing we can do on the blog that's pretty easy is just say, you know, version X or later, or something. So if the the lighthouse and prism one are that like, you're still fine with some people running the old release, I feel like that's probably the right language. otherwise, like, yeah, we can strike the old release. And and yeah, just move on. But those would be my 2 options. Either remove it to like X or later, or just strike the old one and add the new one. Just send me a DM. And I will get the change done. Yeah.

Alex Stokes: Cool. And yeah, otherwise anything else for Petra the fork is next week in a few days. So yeah, exciting. This has been a big one, and I'm at least excited for it to happen. I think I hope most of you are as well. Yeah, anything else. Otherwise. Yeah. we'll move on to the next agenda items. Okay? So then, next we have Fusaka. Again, a little light today on the agenda here with Fusaka. there were some issues with pure dos. Devnet 6, which I think we had said we would plan to base Fussaka devnet 0 on

that being said, yeah, 6 had some issues. And the decision was basically just to crap it. Move to 7 with those bug fixes. Any updates on that. I think people have been quite busy with Petra, so I think it's been a little slow on the period off side.

Barnabas Busa: I could give an update. We we are considering of delaying the possible launch of Devnet 7 till post spectra updates. So once more people could focus on. In the meantime, we have this inside lab, that net up and running. And yeah, ideally, we want clients to work on their fixes like their sync issues. Multiple client teams have indicated that they are currently working on fixing their sync, and they haven't even had time to look into validator custody so ideally from our side, we would prefer, if validator custody and syncing would be the base of Devon 7. So if any client teams would be ready to go with that. Then we could launch. But I feel like everyone is still work in progress for that.

Alex Stokes: Yeah, yeah, I mean, that makes sense in terms of the feature set for Devnet 7. Anything we should discuss now for syncing issues? Or is it more specific to each client and just something they need to work through.

Sean Anderson: So lighthouse has a couple of issues. Jimmy mentioned that it might actually be useful for us to have Devnet 7 running for us to work through debugging. He mentioned. It's a bit easier to work with the F. Panda Ops devnets versus the Sunnyside Devnet. So yeah, some feedback. There.

Barnabas Busa: I mean, we can most definitely launch one. I just wasn't sure if current teams are like working offline on figuring out these syncing issues with like you to assist, or if they're working on, if if it would be useful to have them definitely 7, then the Administration can have it up and running.

Sean Anderson: Yes, for for lighthouse, I would say Devnet 7 would be useful.

Alex Stokes: And Terrence had a question, is there a demo? 7 spec. I believe I've seen a hack, indeed, but I don't know how up to date, it is with, say, the syncing and value or custody changes as part of the spec.

Barnabas Busa: Yeah. So we were considering that shouldn't be part of that 6. And the reason to that was that it's basically a change that wouldn't affect too many things, especially with super nodes around the network. But we also noticed that during the net 6 it might have caused all the issues, because, some clients were reporting that they had a specific subnet. But in the in actuality they didn't have it, or they had struggled serving specific subnets. So yeah, I'm not sure. Everyone seem to say that request is still weeks away. So if that is the case, then maybe we need to have a way to disable the data custody on lines that have it working. Or

I'm I'm not really sure, like. I would only want to include validator custody in this, if we have like, confirmed that that specific client version is actually working and not like a port. the compass implementation.

Terence Tsao: One thing I want. Clarification when we have the spec is the whether to include the new data column site cost by route change. I think, before we just had it as like a single you in 64 column index. But now we change it to a list of index with maps loves for sidecar 1, 28, or something like that. As the cap. So I think that changes things. We need some sort of like a a consensus on whether that should be indefinite 7 or not. So I hope the spec will reflect that. But yeah, we don't have to discuss so much of it for this call. But yeah, I'm hoping to see that in the spec.

Alex Stokes: How critical is that change? For, say, fixing the sink, or any other part.

Terence Tsao: It's not at all but it. But then say, but then say, if today, if one client implements it by other client clear, then they cannot talk.

Alex Stokes: Yeah.

Terence Tsao: Each other.

Alex Stokes: Right? Yeah, okay. so yeah. So in short, there's not a dead net. 7. But that's something we can do very soon. And we can, yeah, be specific. About which of these things go in.

Barnabas Busa: Also do we expect to have a new consensus pack release coming soon?

Justin Traglia: We'll do that the morning or the same day as may not upgrade to like. 2 h after that happens.

Barnabas Busa: And would that also include the fix for the minimal spec stuff.

Justin Traglia: Yeah, that's already been merged. So yeah.

Barnabas Busa: Have all these accounts. Had the time to take a look at that Pr. because it would be great to be able to start testing pure desk with minimal preset in local configs.

Alex Stokes: Which pr, are you talking about? Exactly.

Nico Flaig: Oh, I was.

Alex Stokes: Wasn't this request talking about.

Nico Flaig: Load! Stop! Here! Actually.

Alex Stokes: Yeah, I was asking Barnabas which Pr. Or Justin? Was it the request for payload, or was it something.

Barnabas Busa: Or minimal spec fixing the minimal spec execution payload stuff.

Justin Traglia: Have a good one.

Barnabas Busa: I will be second.

Alex Stokes: Yeah, it's 4, 2, 9, 1.

Justin Traglia: Yes, of it. Yeah, it is.

Alex Stokes: Okay, gotcha. Okay? Cool. So yeah, this is the one they're talking about. And it sounds like. maybe not. Everyone's looked at this. Yet what was the issue here? Just for a summary for everyone here.

Justin Traglia: Oh, okay, I can give a quick summary. Let me look at the Pr. Second. The so the in the consensus specs and the config. We had 3 like requests per payload configuration values, and these are different for the minimal configuration but the El didn't know about this per se, because the Yale doesn't have like a minimal config. So when we change this the Eo would still send more requests than the CEO wanted, you know which is great.

Alex Stokes: Well, so is this a Petra thing or a fursaka thing?

Justin Traglia: This is technically a Petra thing.

Alex Stokes: Okay.

Justin Traglia: Not. I mean, it's just minimal spec.

Alex Stokes: Right, yeah.

Barnabas Busa: Yeah, we haven't done much minimal spec testing before. But yeah.

Alex Stokes: Gotcha. Yeah, no. I thought.

Barnabas Busa: Nice to have.

Alex Stokes: Yeah, yeah, I thought you were just saying, somehow, this is related to the peer dust on that. But I got you now.

Barnabas Busa: Would it be okay for client teams to wait for the this next consensus, back release and then target that for a period of time at 7.

Justin Traglia: We want to do devnet 7 and minimal.

Barnabas Busa: We don't. We don't need to do that for minimum but yeah, I don't. I don't know if there's any any other changes that will hit. In the meantime.

Justin Traglia: Okay. I would also like to get more eyes on the consensus backs. Blob schedule. Pr. I just post. I just shared a link. If more client teams could review this and give approval, I would like to merge it soon. Thank you.

Barnabas Busa: Oh, yeah. Would this be then part of the next release also?

Justin Traglia: For Devnet 7, or it would if we merge it soon, it would be part of the next release. But I'm not sure if it should be part of Devnet 7 specs, because it might not be such a simple change.

Alex Stokes: Yeah, I think it would make sense. I it depends. If we do, it doesn't. 8 or not. But given that, we kind of deprecated 6, I could see doing 7 with the 6 fixes. and then 8 with us, and then from there we go to Fussaka, Devnet 0.

Justin Traglia: Yeah, that's kind of what I want as well.

Alex Stokes: Okay. I also don't think people have the bandwidth to make that call today. but we will pick it up after picture. Let me check. Bernabus. Beta 5 is what we have right now, or that's yeah, have it? I mean.

Barnabas Busa: We have as of 3 weeks ago. I think so. Let me fix was targeting Beta 4.

Justin Traglia: So yeah, beta 5 plus probably the that like the request, Pr, that Terrence was talking about the data column by root requests change.

Alex Stokes: Okay, let's provisionally say that. And we can address this on the next Paradise breakout to dial it in Barnabas, I can. Oh, there we go. Thanks, Justin. Cool. So. assuming there's nothing else super timely for pure dos. Then that was the agenda. Today I'm fine to wrap up early. Everyone is very busy with Petra, and I think it makes sense to focus on that. Yeah, Ben.

Ben Adams: Well, there are 2 other items on the agenda. Yeah.

Alex Stokes: Okay.

Ben Adams: Busaka Cfi. Oh, no, no, sorry I'm looking. I'm looking at the wrong agenda. Don't worry.

Alex Stokes: Gotcha. Yeah. That was last call.

Ben Adams: See ya.

Alex Stokes: Yeah, I mean, that was all I have. Yeah, if there's anything else that we'd like to discuss, that is on topic, we can. We are not going to discuss Uf. Terrence. And otherwise we'll wrap up today. and people can focus on Petra Maina. Let's hope it goes smoothly. So anything else. Otherwise we'll go ahead and wrap up. Okay, happy factor everyone. I'll see you next time.

Sean Anderson: Bye.

Terence Tsao: Can't beat my future.

Fredrik Svantes: Bye, thanks, bye.

Justin Traglia: Bye, everyone.

Ansgar Dietrichs: Bye. Thank you.

Pooja Ranjan: You know.

Transcript adapted from ethereum/pm, CC BY-SA 3.0.