Merge Community Call #7

Transcript

welcome everybody to the seventh and final merge community call um let's see what's happened since the last one quite a bit um bellatrix obviously happened on the sixth uh and that's basically the final final thing we were waiting on before mainnet merge um same is there anything major uh why don't you just start and i will make sure this link is right in all the places sure um yeah so good morning or afternoon everyone um yeah like currently saying this is gonna be the last of these community calls hopefully uh under something i went it's really bad with emerge uh but yeah i i think we

were safe that's the last one um i guess there's two things that i think are worth highlighting today um the first is that uh yes we had bellatrix uh go live yesterday and we'll have the actual transition happen uh sometime next week uh in order to track that uh there's a bunch of different websites i've personally been using bordell.wtf to track it um so because we use the value we we we can't have a specific time for when the merge happens but it's roughly looking like it'll be late uh utc on on wednesday um since uh

first announcing kind of the the actual uh ttd and and merge releases a lot of the execution layer client teams have put out new releases um that uh either fix some bugs or have just some suggested improvements um so if you are writing a node it's worth having one more look at the blog post and and seeing if you're on the latest version of your client um and if not you can kind of look at the release notes on the client you're using and and determine whether you want to upgrade um so this is mostly for the consensus sort of mostly for the execution layer clients um so yes please have a look at the post uh

at the post itself and and kind of make that call um and then thirdly uh this morning uh literally minutes before this call we put out another blog post about the test net shutdown um so several months ago we announced that uh killed robsten and rinkeby are deprecated with the merge and kill was scheduled to be shut down when the merge happens and so this is what will happen uh so next sometime next week uh that test net will be shut down um obviously it's kind of you know impossible to like fully shut down a test net anyone can still run a note on it if they want so by shutdown what we mean is kind of the nodes the

boot nodes run by client and testing teams will be turned off um and then infrastructure providers will likely stop supporting uh that network as well and so you know you can still run a node but it might be much harder and if you're an application or user who just wants a test net to use that's a smooth experience um that probably won't be the case on those um so kiln is going to be sunset next week and then roxanne and ruffson is scheduled to be shut down by the end of this year and ring could be sometime in the first half of next year um that said uh we've heard already from like several infrastructure uh providers that uh they're gonna stop supporting or kind of gracefully

shut down um support for robson and rinkeby so that might also mean just less functionality and stuff like that um so if you're using any of kale and robson and rankibi um i highly suggest you move to gordie or sepolia those two are are very much intended to be supported uh long into the future um and hello to all the mcon people traveling in it's pretty neat to see the room they're slowly filling up hey um yeah so those are those are the three things i had um oh and yeah i guess actually one last uh quick last one if you do want to watch the merge happen uh eat steak here bank class cat herders and a bunch of others are organizing a community call livestream when the merge

is happening um i believe this will be on streamed on the ethereum foundation youtube but they're all the ones kind of um uh hosting this and and facilitating this um so i'll try and pull up the link after but yes if you want to watch it happen um that's probably gonna be the the main place um yeah that's that's all i had yeah there's nothing else that comes to mind oh thanks puja for the link yeah who just think is that the livestream um but yeah bordell another time uh double check your el client versions if you're

using kill and rob center and could be the time to move away and if you want to watch live stream um the link that pooja shared is the right one so yeah this is the time for any questions and if we need to stall while people type them out happy to make space for client teams or researchers if they want to mention anything specific we may have forgotten or anything new that's come up recently so if you're a client team or someone who would like to make a comment from the research side of things just raise your hand and we can unmute you

talk about the latest shadow fort please yeah i think it went uh off without a hitch there weren't any issues uh perry is here he's welcome to give a summary i don't think perry is here yeah i don't think there was much beyond that other than it went well which is uh it's a good thing to hear yep it seems like this might be a pretty

short call again if people have questions just put them in the chat we're not going to be talking this entire time um so we're happy to answer questions if they materialize but we're not going to sit here for an hour if nobody has any questions or anything oh there's someone from emconn can ask a question very teacher's message with the is anyone scared i'm i'm not scared

um no yeah i'll say you know it's very possible that we see some minor issues when the merge happens like you know we saw in bellatrix there was like a a short drop in participation rates i think we expect those things to happen and and we'll we'll deal with them as they come and yeah like the way a theorem is designed like it's it's made to operate in very imperfect conditions and so i think um we don't need 100 of nodes to update exactly perfectly for this to work um obviously we want to get us as high as we can um but if we see and i suspect we'll see some minor issues on the day of you know myself and others will kind of chase those down and and make sure they get fixed but yeah like

we did on bellatrix and and bellatrix is basically back to the pre uh sorry like we did on the beacon chain and and the beacon chains basically like the pre-bellatrix levels of at the stations and participation so yeah i'm i'm feeling good yeah i'll also share given we have a little bit of extra space here check out that link i just put in the chat it's a compilation of a bunch of responses from core devs including their uh summary of their feelings leading up to the merch and you can find out if people are scared in those comments okay so there's a good question about exchanges and mining pools and whatnot and like ensuring user safety during the

merge um and so for mining pools it's it's obviously like a bit different because like i'm not a my pool operator um but i suspect what i would do if i was one is something like you want to mine all the way up to the last block and you know distribute those rewards um there is like some potential for uh reorg on the last proof of work block until the proof of stake chain has finalized so that might require like some just monitoring on the order of minutes um and as i understand it like mining pools don't send out payouts instantly they usually like bundle them and what not um so i think if you're a mining pool mining all the

way up to like the last proof of work block making sure that like the chain has finalized and has come to consensus on that last proof of work block before sending out your rewards is a smart thing to do um similarly for an exchange you know exchanges do this every network upgrade but they typically pause withdrawals and deposits just to make sure that the chain has settled um so if again i was operating in exchange i would do this you know shortly before the merge up to the point where it's it's finalized on the other side and and you see that the chain is stable um beyond that you know there's nothing that like mining pools or exchanges should have to do except

that's different than what they do in any other network upgrade which is you know upgrade their nose and monitor the chain stability and and once everything is kind of clear then uh they can they can reopen uh deposits and withdrawals um but yeah during that time there might you know if any issue come up you'd much rather have deposits of withdrawals uh halted um and then for non-censoring any boost relays uh there's an answer in the chat around blocks route uh manifold as well said they they were working on one but it's unclear if it's if it's released yet um so there's some links in the chat there um risks if risks

uh risks if the hash rate stops to go down uh the hash rate starts to go down so basically if it goes down a little bit um not much happens you know the merge just happens later than expected we reach the terminal total difficulty later um over the past handful of days it's actually gone up a bit so um but it might it might go down in the next couple days um you know so if we reach the merge on like the 17th rather than the 14th we'll do nothing we'll just wait an extra three days um if there was a world where say for whatever reason 50 of the hash rate drops overnight um which is very unlikely um or you know it it's just like 10 a day over the next week or

something like that it just makes it that instead of reaching ttd on like the 15th or 17 were like you know say a month out um then we could coordinate a ttd override um which is like we did on some of the test nets where we just choose a lower ttd value and put out new client releases with it um and that coordination you know choosing the value putting out the releases getting everyone to upgrade usually takes at least like a week so it's not worth doing if the hash rate is not dropped enough to delay things by over a week um but if we saw that it's dropped enough to delay things by say a month especially in the case where like it dropped below like 50 or something and there's maybe these weird scenarios where there could be like a 51 percent

attack if that hash rate came back online um client teams would probably decide to just change the tpd and and do an override um but it's not looking like that's that's going to happen um mika you have a comment on that yeah just in order for at this point in order for us to be delayed by a month the hash rate f drop by like 90 or something right yeah that'd be very very significant now because and the closer we get to that date the more it has to drop so like tomorrow it'll have to drop 95 or something yeah that's right that's right yeah given over like five days away or something it would have to drop um um do you think there will be a big fight infrared by users running their own el um

i don't see why so basically infera does not support the engine api um so that means that if you use infera post merge as your el uh you won't be able to produce a block and and to verify blocks that are being produced by the networks um so it would be kind of strictly worse in terms of like potential rewards um obviously it might be lower cost to do that but i it's not something infrared has like mentioned to us they've also like done a ton of education towards their users over the past couple months trying to explain to them why they uh why people need to run their own el now

um oh yeah okay it seems like i misunderstood the question what was the question yeah i believe i believe the question is is it is now significantly more expensive to run an ethereum client than it was yes after after regular node operators um it had will become much more expensive and the reason for this is uh twofold when you have to run two clients is more operationally complex but two you need to retain receipt history and i believe block header history and i maybe block body which i don't know on that one whereas previously you could run like a number of the different clients allowed you to prune history in order to save space so you could run it on like 150 gigabytes of disk right now you need like six or seven hundred

gigabytes for the i think the smallest node requires like six hundred dollars now okay okay so there are a number of people who previously could run a node and now cannot run a node um and so there's discussion on maybe trying to fix this but there's no eta on on fixing it um but yeah i i personally do suspect that there will be some number of users i've seen people frustrated with the change and some people giving up and running nodes because of this yeah so i think like yeah so short answer is like of course if the requirements increase less people will run the nodes um and the questions like by how much um i think you know client teams are working on ways to like optimize that i know you know the get team and i believe

a while back eragon was working on a way to like run you know a beacon like client in parallel and and i know that the cl teams have worked on ways to like um optimize the storage as well so i suspect over time like the the overhead of running both is something that will work on minimizing but for sure in the meantime in the meantime if there's more overhead then it means less people might be able to run this um and so they might use something like or alchemy unfortunately yeah um okay uh next one was uh has a difficulty bomb some side effect um so i'm not sure i understand the question

but uh basically the difficulty bomb is probably starting to go off right around now um i haven't looked at it um super closely myself but it's it's very much in the early stages uh so you know there's probably some minimal effect um and if i don't know if it's showing up in block times yet but it might show up and say mining rewards um every hundred thousand blocks it increases um so and the merge i suspect is going to happen last i check it was like something like 40 000 blocks at the current difficulty um so wherever the difficulty bomb is kind of at now um it's likely that it will it won't go more than like one extra period beyond this before we we

transition to proof of stake um so there is some like minimal effect uh it might grow a little bit more um but we're not in a spot where like the difficulty bombs impact uh in the next week or two will make a difference to things okay so there's some discussion in the chat there's some questions in the chat about about uh basically this history storage and whatnot a question about what's the lightest weight non-validating home node setup um i don't have the exact answer here i don't know if someone on the call has an answer oh i see perry's joined on the call

perry i feel like i might you might have a feel for that oh okay sorry i've muted you sorry about that yeah sorry um yeah could you just repeat the question again if you have a feeling for what is the lightest weight non-validating node you can run um if you're optimizing for like this space basically i think either nether mind and i think netherland and nimble should be a good combination in that scenario i remember talking to malek about it at some point and i think the difference isn't that much with geth

but it is still you're still gonna save a couple gigabytes do you have a ballpark numbers or not off the double [Music] let me try and find that i can just post them in chat in a few minutes awesome yeah thanks we had some people also ask about the shadow the latest shadow board um other than saying it went well do you have anything to add there barry uh yes i think yeah so uh yeah it it did go well um the only issues we saw was with one teku aragon note but that was my mistake i

didn't clean the machine properly earlier so it had some stale data and aragon basically thought it was on the old shadow fork rather than the new one whereas the beacon no thought it was on a new one and it just was unhealthy um besides that we're just looking into uh some mev relay behavior just to make sure that everything is working as expected but that's the only open topic right now i think yeah otherwise all clients went through a transition as expected thank you um fred has a question what's the best

public dashboard to follow during the merge um so i think unless you're like monitoring your own nodes um the first thing you probably like beacon chain has a bunch of charts they usually lag a little bit unfortunately but i think that's probably the best public one um on the live stream i suspect people will be sharing their screens and looking at like more granular monitoring as well so um the live stream we mentioned earlier we'll probably have some better uh or like more uh recent data um and um and basically i think the first thing you were looking at is like does the network finalize on the other side and that basically tells us you know over

two-thirds of the validators agreed on on uh the merge block and the transition and we're coming checking along that's a very low bar for success um i think the second thing we look at this participation rate so it's like you know between two thirds and a hundred percent how many validators are still uh are still doing all their duties um and i think from that you know there's a bunch of other kind of secondary metrics we'll start looking at um one i'm particularly curious to see myself is like fee recipients you know uh are there cases where we can we can tell that like large validator sets have not or or miss set their fee recipient um i think you know from there it it becomes

um you know it becomes much more cases where like you know the network is coming is going along we haven't lost any throughput and whatnot and it's more about like are there specific operators that that are having issues or or do we see you know we might see some issue with a specific client combo and trying to to try and grade those um but yeah beacon chain has some some good dashboards that will give you this high level information um if you run monitoring on your node you get a bunch of other kind of health data and just stability data and um and and yeah i think on the livestream we'll be sharing some more stuff as well um do we plan to post subtitles for this meeting later where if yes um advice subtitles i think youtube like does the closed captions automatically

uh when we after the live stream is over um i don't know if we do notes for these meetings if we do full transcripts uh i think yeah i don't think we do yeah oh we do just as we do so there'll be a full transcript as well um usually the transcripts lag by a couple days so um if you if you want to get the info before the merge i recommend in this case actually watching the recording uh because you might get the transcripts like next monday which might be like 24 hours or something before the merge any other questions and perry is posting all of the the size of the different kind themes in the chat

right here yeah if there's nothing else oh here we go how many validators do you think will run lbv boost after the merge um i don't know um i know some sticky pools have like made commitments to run specific um like specific mev instances and and so

i i suspect you know they'll probably be a couple of like the larger chunks of validators that end up running it um but i don't know for all the staking pools i don't have a view and say like what exchanges will do um and it's hard to predict what like the long tail of home stickers and solo stickers will do as well um i don't know if any if anyone thinks they have a better estimate oh i see i see there's like a chris here but i don't know if that's chris from the flashbacks team or um if so please post in the chat and all on youtube okay there's a question about writing a

validator note on the raspberry pi i don't know perry if you want to answer this um sorry i muted you again yeah no problem um yes so you can run a validating node on raspberry pi we have the there's a twitter page theorem on arm so they pro they um they prepare images which have the clients built in in them and i think they even have a couple of guides on how you can get validators running on raspberry pi

they have taken part in shadow forks in the past and everything's work quite okay i think the only thing they recommend is that you don't run too many validators but if you just have a handful it should be fine and they were testing a different board rocks something if i'm not wrong um it's supposed to be a lot more powerful overboard compared to um raspberry pi and it had some different architecture that allowed you to have better ssds connected to them i know there would like a lot of advantages to using that one um so talk to them they should be the best people to help with this yeah i see something in chat rock by something like that yeah

um yeah it's pal question um i don't have strong i guess you know the only strong opinion i have about these is that um if you're a user uh you should try and figure out whether like the chain you're interacting us um is is is safe in a way and like there's many ways you can look at that and there's many things that people forking ethereum can do uh in order to make things safer um and basically you know one of the the very obvious ones is just like replay protection like if you make a transaction on a chain you want it um you want it to

ideally not be able to be replayed on another network so that you know the transaction you're only making one transaction when you think you're making one transaction uh to me personally that's like the very minimal bar of like safety um and you know some of these chains like these these potential forex have not um then this this like very basic technical thing so you know my thoughts on it is like none of these forks have released software that's at the very minimum bar of secure for users um doesn't mean that they can't do it doesn't mean that like they're not free to fork the network um but yeah that's my rough feeling is like so far everything we've seen from them

has been like very subpar in terms of quality i personally would not like trust my funds on any of those chains um and unless you're like very comfortable with what you're doing trying to say like make transactions on one chain and um you know to say like sell your east pal to something else um at the risk of potentially getting replayed on my net might not be worth it um so yeah that's the main thing i would say it's just like you know be sure that like the chain you're interacting with is safe and that it doesn't put your funds on the ethereum net at risk um by doing so um but you know yeah like like i said i people are free to fork the network if

you want um yeah um i can try to unmute i've asked the unused m-con but i think somebody's gonna have to like oh accept oh hey we just want to say thank you [Applause] thanks a lot this is awesome thank you any other questions because otherwise

this feels like the highest point at which we can end this go out and applause yeah it's not it's not gonna get better oh okay uh after how many epochs will you feel comfortable everything is fine uh so like yeah basically we have one finalized i bought like yeah so that's usually two by default but if we were in like if we were in a state where the network was unstable it might be more than two and um yeah so i don't want people to use two as like a number but finalization is is the the bar that we're looking for again though i'll emphasize this is like the very minimal bar um you know if

things are not finalizing that's very bad but even if they are you know if we have eighty percent of the network uh um that's that's that's participating um you know then it still means the network is finalizing moving along but it means we've lost 20 percent of throughput so that's still something like we're not going to call it a day at that point we want to fix that but the network would be would be stable right yeah right yeah 20 percent it's not the end of the world if there's no no but yeah to get to epochs just to set expectations for uh if people are watching the stream

and all of a sudden there's you know another additional epoch beyond two or something no sorry so those are two different things so two epochs should happen if we have at least two thirds of clients updated and and participating as expected um and so that's that's kind of the very minimal bar means the merge has happened and we're we're kind of finalizing on the network but it could mean that we have up to a third of the network that's not operating correctly and obviously we'd want to figure out what the issue is is there and fix it so you know it could be a broken client it could be something like you know some major staking provider has an issue or like several of them have an

issue um so things things like that um yep yeah okay more questions um okay so yeah yeah so uh there's a question about um basically block times and how it impacts protocols that might use time uh that might use uh this to calculate uh annual interest rates so the thing is um on proof of work um timestamps are very like malleable so sometimes protocols don't use the time to calculate annual interest rates instead they use like the number of blocks they make an assumption about you know how many number of blocks there's in a year and they kind of go from there um so if we just very naively

go from 13 second to 12 seconds it means there's more blocks in a single year um so if you've been using that um if if you've been using that uh kind of proxy of like you know a year divided by 13 seconds to count the number of blocks of the year now you probably want to use 12. none of those are like perfect proxies because the 13 seconds you know we have stuff like the difficulty bomb that shows up um and it's it's only like a rough average and then the 12 second we have things like uh validators being offline which could which could cause some slots to be missed we see this less than like uh less than one percent of the time right now um so you know if you are using number of blocks as a proxy for time you should

double check those calculations um and the thing that's also worth noting is that post merged timestamps are actually a reliable indicator of time so the time stamps will indicate the time stamps on the execution layer indicate kind of the start time of the slot um so if you just want to like straight up use time for your calculation um you can do that in a more reliable way post merge um okay and then there's an article about uh about more details around network finalization um so okay any particular hack concerns closer to the merge events um so the thing the thing with the merge is basically like the execution layer itself doesn't really change much right

there's one op code that changes um so in terms of like smart contracts being hackable there's not really a ton that you're you're introducing here like um it's not like we're adding some new up codes and they can be used in the way that was previously unusable and kind of affect the the outcome of of transactions um and that's by design right like you don't want the merge to require upgrades from applications so ideally you shouldn't change the behavior on applications um so i think at that level like that's not something we're concerned about um you know i with regards to stuff like bridges and oracles and whatnot that kind of feed in to like you know that like

report data back to the chain um that becomes very much like an infrastructure a great question again where like you know um oracles should also upgrade their nodes and they should make sure they're on the right chain um but that's something they must do every every hard fork right um yeah so i don't i if somebody else has something to share their uh post in the chat and will unmute you but i don't think there's anything new on like the smart contract hack side um yeah so so yeah there's a question about

like what happens if we don't have two thirds and we're not finalizing um so like yeah you basically not like you you you don't have a strong guarantee that like the chain you're on wall pre-org as even more so if if yeah your view of the network drops below 50 um so if you are say like an exchange or something or someone none even just a user you know who's settling in an off-chain transaction or selling a transaction which has some off-chain implications you just can't be sure that this won't reorg whereas once once a block or once an epoch gets finalized um all the transactions within that epoch um are basically extremely unlikely to ever be reorg um

and doing so would require burning about a third of the stake at a minimum um so unless you're transacting in amounts that involve more than about four million eth um the cost of reorganic finalized blocks is much greater than whatever you're selling um okay barry there's some people who want more clarification on shadow forks on youtube again just more you wanna add um we're still digging into it so i'm not particularly sure so we just noticed that uh there were a couple of teku blocks that had missed their proposal because um because they were connected to a relay and the relay was offline

but it had already accepted the bid and that shouldn't have happened so we're just checking us what's happened or what the assumption is that's happened there um but we're still figuring that one out it looks like everyone's behaving as expected though so it could just be that the proposal was delayed or something i can share some more info once we have it anything else

okay last call okay i think i think that's a good spot to wrap up then um yeah thanks everyone uh for joining and yeah see you on post marriage ethereum in less than a week thanks everyone emcon we see you clapping and hear it yes yeah all right see ya okay have a good one cheers