All Core Devs - Testing #68

Transcript

Barnabas Busa: No.

Pooja Ranjan: Hello, hello.

Justin Traglia: Good morning.

Emmanuel Nalepa: Good morning.

Justin Traglia: Good morning, POTUS. I wasn't sure if you were gonna be here.

Potuz: I'm just in the middle of nowhere. I'm not sure if this is going to last, but for now it's working.

Justin Traglia: Okay, and I can hear you just fine, by the way. For the middle of nowhere.

Barnabas Busa: I can hear you better than usually.

Justin Traglia: Yeah, no joke.

Barnabas Busa: You probably have better internet there in the middle of nowhere than in the office.

Justin Traglia: Yeah. But nice, yeah, it does look like the middle of nowhere. Middle of a jungle.

Barnabas Busa: I'll just give it another minute, and then we can get started. Welcome, everybody, to All Core Devs testing call. Pooja, can you… Get the streamer started?

Pooja Ranjan: We are live.

Barnabas Busa: Thank you. Welcome, everybody, to All Core Devs listening call, ACD number 68. It's the 2nd of February. And, let's get right into it. So, the agenda is quite busy for today. First of all, let's get into Fusaka. We have been working on the partial cell proofs in the past couple of weeks. We have launched the Blob.net Zero, and currently our supernodes are struggling to keep up with the, blob demand, because, they've adjusted the minimum epochs for data column sidecar requests to 256, so that means that the number of blobs that we should store is, up till, only one day, but instead, Prism and Lighthouse Do not seem to be respecting this flag, so, we're gonna need to…

either do something to make sure that they will respect it, or look the databases, or adjust the number of blobs we send through. Otherwise, this… that note is going to get very expensive very fast. We can probably discuss the async, but, It's just something that I wanted to raise today. Does anyone have anything to say regarding Blob.netzero? Daniel, commented that he's gonna take a look into this, and hopefully someone from Prysm can also take a look. And we are, looking for other CLs to also, Begin implementing the partial cells, or finish implementing the partial cells, so that we can add them to this template. Thank you, Venice. Okay, next topic is gas limit testing.

Would anybody have any updates regarding, current, benchmarking? Mmm, maybe… This one? Or… I don't want to put anyone on the spot. I'm not sure if we even have any updates. In case we don't know, I will propose that, we will, have, Rafael from our team, come online next week, to showcase benchmarking tool, benchmarker, that is going to help, the ELs basically benchmark their EL performance. Hopefully he can give us a short, 5-10 minute live demo, and, show us how it works.

If there's nothing else regarding estimate testing this week, then we can quickly move on to the next topic already. Which would be Glamsterdam? Regarding, ball W2, stefan, could you give us an update?

Stefan Starflinger: Yeah, sure. So, I've been testing, over the weekend, all of the different clients. I will share the spec, in the chat. And I try to keep it up to date. It might not be 100% accurate everywhere. And, I shared, some issues also. In the, ethereum… I'll share it here also. So in general, most clients are pretty far. There has been a test release, 5.1.0, maybe not all clients are aware of that yet. there are still a few clients that need to remove the gas parameter from the receipt, so Bezo and Nethermind still need to remove it. This is from the first, so if you push anything today, then it might not be 100% up-to-date. There's also some other issue, I think, that's already been fixed in NetherMind, with the get payload. Then Nimbus, I identified the issue that still needs to be fixed, where, the activation

of some Gloas features is already happening too early, in Fusaka instead of in Gloas. It's described here as well, and then Geth is already working on some issues here. And I also fixed some small issue in Lighthouse, so we have both Lighthouse and Lodestar SCL ready for testing. In the spec sheet, I have a small example for a Kurtosis config, and if your client is working really well, you should also try to add the EVM fuzz. to the network and see if it, can, work under some transaction load as well. So, PK has already modified the latest spammer to… and the EVM first scenario to include the new opcodes. So, slot num op codes and the other call stack opcodes as well. So that would be great if you have some time to test with the spammer as well.

I think that's pretty much it, and the network will launch on Wednesday. And I think, we should be able to get everything ready by then. It would also be good to get an update from clients, as well, but I'll hand it back to Barnabas.

Barnabas Busa: Yeah, so just before the client updates, maybe we can still discuss the two open, PRs that are 726 that I just linked. This is an execution API PR that adds, let's use an RPC method. Maybe, can you get an update on this?

Stefan Starflinger: These are not a blocker. It would be nice to include them in DevNet2, but we don't need to have them ready for the launch.

Barnabas Busa: I, yeah, I understand, I'm just curious if there's any blockers in merging these in.

Stefan Starflinger: I'm not aware of any. But we can ask Tony if there's any update on that. I'm not sure if he's on the call.

Barnabas Busa: can see Tony on the call, so maybe we can check with him. No, I'm on the call. What was the concrete question? Whether we could, get 726 and 727, merged in.

Toni Wahrstätter: Okay, yeah, I will have a look, and let's… Figure that out. Async.

Barnabas Busa: Yep, sounds good. Let's, target the mergers in… by Wednesday as well, if you can. Okay, so can we do a.

Toni Wahrstätter: Alright, one of them is the JSON RPC. Would be great if clients can give input if they have implemented that already. I tested your, Cortosis config, Stefan, and it seemed like at least for GEF, the JSON RPC didn't work yet. So it'll be…

Stefan Starflinger: I don't think it's implemented by any client yet.

Toni Wahrstätter: But I think it's.

Stefan Starflinger: shouldn't the spec be merged first before? Or generally, I don't know exactly what the best process is here, but generally, I would assume that first the spec should be merged, and then clients implement. For it.

Toni Wahrstätter: Right, yeah, it would be great in this case, if we get this back, if we get the PRs merged.

Barnabas Busa: Yep.

Toni Wahrstätter: One of them is engine API related, and the other one is the JSON RPC methods, and both of them are approved, and, yeah, I don't see any reason why we shouldn't merge them now.

Barnabas Busa: Yes, Jared, that's what you're talking about. Would any clients have an update, whether they have taken a look into implementing these two, PRs, and whether there are any blockers? Maybe you can go in order. Anyone from Besu?

Karim Taam: So, Besu, we are close to be ready. The engine API, we merged the two engine APIs. We still have some open PR, so… We have an open PR for metrics, we want to add some metrics for block access. We have also the prefetch, it's open, but I think we'll try to merge everything Today or tomorrow, and it'll be okay. So… just, some PR2 merge. We are passing some… the test we tried with the last version. And we also did some tests with the local testnet, so… yeah, for now, it seems that we are okay.

Barnabas Busa: Thank you. Anyone from Nethermind?

Marc Harvey-Hill: Sorry, yeah, so was this the, engine API method to get the valves directly? Because I… I thought that that was changed, so it was just included in the payload.

Stefan Starflinger: I mean, there's a… there's a payload, there's an engine API payload part, or the GET payload, and we have the engine fork choice update version 4. But the methods we're, describing here, I think these are… The, engine get payload bodies by hash, and, get payload bodies by range.

Marc Harvey-Hill: Yeah, yeah, so, I implemented get block access list by hash. But I haven't yet made the change to move it to the payload.

Barnabas Busa: Right. Anyone from GEF?

Jared Wasinger: Yeah, we're… well, I am still fixing bugs. Broken blockchain tests, I still need to implement the, the Edge and API method, but, Yeah, planning on trying to knock that out and get us ready in the next few days.

Barnabas Busa: Okay, thank you. From Russ?

Jen: Let me take a look and give you a proper update. We have some PRs in progress. Can you clarify when you need these by? Are they… are they important for DEFNET2?

Barnabas Busa: Ideally, the PR should be merged in, either today or tomorrow, and then we should target these four that include, yes?

Jen: Okay. So by Wednesday, you should be able to do that.

Barnabas Busa: Thank you. Ergon? I guess somebody from Erigon around, Okay, is there any other block-level access list topics that we need to discuss? everybody, should be aware of all the bugs that are Prysm in their clients right now, and we should be able to target the DevNet launch by Wednesday.

Stefan Starflinger: Yeah, it would be good if we could get also the BAL tests numbers up. That would be great, and if there's any blocker to let us know.

Barnabas Busa: Caleb, is that Nimbus DL? Or block number access list?

Caleb: Yes, it's Ace. So, basically, the invoice is pretty much ready to go. I don't know, one thing to notice that… So we… we treat Glamsterdam as the relevant folk, instead of the… the AIP featured him in the bow, so basically… That's it.

Barnabas Busa: Right, but this is on the CR side, right?

Caleb: Yeah, yeah, on the CL side, yeah.

Barnabas Busa: Okay. Yeah, we can give this a test as well. Stefan, if you… So the bandwidth to add, bandwidth also into the testing scope?

Stefan Starflinger: Sounds great. I will have a look.

Barnabas Busa: Okay, thank you. Next topic is APBS. it'd be good if we could get a… Update from the different client teams regarding their, Implementation? progressed?

Stefan Bratanov: I can give an update from Teco. Basically, implementing, now, the… New specs, which will be released, hopefully, today or tomorrow. Yeah, that's… that's pretty much it, and then we'll focus on the… on the rest for DevNet Zero. We still have quite a bit…

Barnabas Busa: We do…

Justin Traglia: need to decide if this is what we want to do. Like, I need some confirmation.

Stefan Bratanov: Oh, okay.

Justin Traglia: Yep. I guess this ties into the…

Barnabas Busa: Alright, moving. Before we go into that, is there any reason that you guys are looking into implementing the next spec that hasn't even been released, instead of focusing on the scope of zero?

Justin Traglia: It's because we're talking about, changing the scope of DevNet Zero. Like, clients want the… The change, which includes moving the, block KZG commitments To the, to the bid.

Barnabas Busa: Okay, so the idea is to basically have the Alpha 2 as the scope for 00.

Justin Traglia: Yes, that's correct. I could, I mean, if we agreed to this, I can make the release, like, right now, and it would be ready, this time tomorrow.

Potuz: There's no program inside of Prism, either way works for us.

Eitan Seri-Levi: White House.

Barnabas Busa: from an iPhone? Yep.

Eitan Seri-Levi: Yeah, Lighthouse prefers that we do the blob KCG commitment stuff for DevNet Zero. But yeah, we were making steady progress towards, like, an end of February DevNet. And, yeah, we have those, data column sidecar changes, like, moving the KZG commitments to the payload bid ready, And, like, made a lot of, changes that are ready to be merged around the new spec. We're also passing spec tests for Alpha-1, yeah.

Barnabas Busa: Are there any other changes that, could basically slow us down. I know that everybody wants to basically have this…

Potuz: These changes do not slow us down, though. These changes actually would make it faster for us, because it simplifies a lot on the implementation side. And all of the remaining changes on the spec, as far as I know, they are not hard work, so we can just start merging them as soon as we want. And we can… we can or may not deploy them on the definite, that's fine.

Barnabas Busa: Okay, so anyone from Lodestar could also chime in?

NC: Yeah, we're in favor of doing it.

Barnabas Busa: Nimbus?

Caleb: Hi. I'm sorry, are we talking about, PBS.netzero?

Barnabas Busa: Yes, we were talking about the scope of EPBS.0, and whether or not we should be targeting alpha 1.7.0 alpha 2, or Alpha 1 as the target, consistent spec release.

Caleb: So I think we should be targeting the… There are offers at 1.7.2, because, We've pretty much implemented all the changes from the previous release, and I don't see us spending so much time in implementing anything new, actually, so I feel like we are good to go anyways.

Barnabas Busa: Okay, so in favor to do the new respect.

Caleb: Yes, yes, we know, yeah.

Saulius Grigaitis: Mmm.

Barnabas Busa: Anybody from Grandine?

Saulius Grigaitis: Yeah, so I think that the later spec, it's better, because, I mean, we are a bit far from, I mean, like, a month or so until the DevNet, and it makes sense to implement now the the newer spec. So yeah, we support the latest spec.

Barnabas Busa: Alright. One more question, regarding the Alpha 2 is the variable PTC deadline. Justin, was that the one that, was… outdated?

Justin Traglia: I don't… I don't plan on including that change in this release. We can save that for the next release.

Barnabas Busa: Right, okay.

Potuz: I'm sorry, we change?

Justin Traglia: Francesco's variable BTC Deadline PR.

Potuz: Just… just in case this is not clear, when people say, like, we're targeting this or that spec. That's not really what's gonna happen on the DevNet. The DevNet is going to have the structure of those specs, but not the feature of those specs. So, we're going to target a strict subset of the spec.

Justin Traglia: Indeed.

Potuz: So, those changes, you can or cannot merge them, because it's fine. I mean, the PR… Francesco's PR, as it is now, it's not mergeable, but whenever we merge it, it's fine if we do it or not, because it will not affect any DevNet.

Barnabas Busa: Also?

Potuz: for choice is kind of irrelevant for DevNets. We're not testing for choice.

Barnabas Busa: But wouldn't the variable PTC deadline be,

Potuz: We are not going to even send PTC attestation, so it's fine. It doesn't matter. We're gonna build on top of the full block. Like, dealing with the subtleties of timings and when we're gonna reorg and these kind of things, this is going to become much, much later in the testing scenario.

Barnabas Busa: Right, okay, let's, push the variable PTC deadline, PR then out of the scope of Alpha 2. And, target that for a later release. Trustin?

Justin Traglia: Yep, that sounds good to me. I will make a release after… after the call.

Barnabas Busa: Sounds good. And then, regarding, implementation readiness. Can we still aim to launch around the 18th of February? Is it, does it sound reasonable? And with the new specs, and that would be Alpha 2 released then? I guess, nobody's quite sure yet whether implementations would be ready by then, so maybe we can push that question to next week as well.

Is there any other updates, regarding… Epbs topic?

Justin Traglia: Nothing from my side. Or actually, wait, we did close the, remove min builder withdrawability delay and, replace it with the one that lowered the value. I think we chose 64 epochs. Or… Something like that, yeah, I think 64.

Barnabas Busa: Okay, then that will also be part of the Alpha 2?

Justin Traglia: Yes, I will. But it'll be such a… such a simple change, just changing it. Constant, or config. Yes, that did merge protos. I just shared the PR.

Barnabas Busa: Thank you. Okay, if there's no more Glamsterdam topics, then we can move on to Hegota, as there's not that much happening in Hegota yet. Just one point to note is that the headliner proposal deadline is approaching. There's only two more days left, so if you have a Hegota proposal, headliner proposal, then make sure to submit it in the next two days. And it looks like, actually went through… Oh. Portos… Had a comment as well? But I think we discussed that already. And we discussed all the ball topics as well, so…

Does anybody else has anything else to add? It's not, that's it for today, and let's get back to work. Thanks for coming.

Potuz: Nice. Bye. See you guys.

Justin Traglia: Bye, everyone.

Marius van der Wijden: But…

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