Everything EOS Mandel Launch Group, 6 Jul 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: EOS Mandel Weekly Meeting #22 (July 6th 2022)

Description

This week the Mandel Huddle group updates us on RC2 development, Ubuntu 18.04, FILLPG support, and software release schedules.
Summary

🟢 In progress action items are on track:
#41

RC2 will be released today, but still requires testing before we can activate the consensus features.
RC2 includes improved support and documentation for Ubuntu 18.04
CDT and contracts will be released on Thursday
ENF is looking for feedback: is FillPG something that should be updated and maintained? EOS Support will prepare a survey.
#55

Full D.U.N.E. 1.0 release, including eosio::tester and support for RC2 will be released Friday
We've received additional feedback on D.U.N.E. from respondents of the EOS Support survey:
DUNE_Feedback.pdf
Survey indicates there's an opportunity to improve awareness and correct misconceptions with a D.U.N.E. marketing push targeted to smart contract developers
#44

There was a no-go decision for feature activation this week. We will revisit next week.
BPs are asked to upgrade and test RC2 this week - in particular, we need to validate we have a working history solution before next week's go/no-go decision.
A number of BPs have upgraded successfully to Mandel 3.1 RC1.
#40

13/15 BPs have signed the MSIG on Kylin to update EOSIO permissions granting authority to EOS Nation and EOS USA to facilitate more rapid updates in preparation for testing.
Participants (23)
Host: Daniel Keyes | EOS Nation
Matt Witherspoon
Bucky Kittinger
cmadh | liquiid.io | Corvin Meyer auf der Heide (Corvin Meyer auf der Heide)
Kevin Heifner
Max_Cho EdenOnEOS Member-South Korea
Stephen Diesel
Andrii Cryptolions (Andriu Bobrovych)
Ross (EOSphere)
Ted Cahall
Jeff Werner
Brendon Ross
Randall Roland
Genie Kim - EOSeoul# NEOPLY
swatanabe
EOSUSA Michael
Zack Gall
Alysha Mims
junhank | NodeONE
Vlad
Johnsonb - Object Computing# Inc.
Denis Carriere
Rob - Genereos

Timestamps
00:00:00 Start
00:00:02 Last weeks recap by Daniel
00:00:32 Development updates and RC2 testing by Ted
00:05:40 Go no go and Hyperion by Daniel, Michael, Ross
00:09:30 Jungle testing and Improved support for Ubuntu 18.04
00:15:30 FILLPG support and Dfuse by Daniel, Michael, Corvin
00:26:40 Platform support and survey feedback by Daniel, Bucky
00:29:47 Software development release schedules by daniel, Ted
00:32:20 Wrapping it up a meeting overview by Daniel
00:34:23 End

Everything EOS is the official media channel of the EOS Network Foundation (ENF). The EOS Network Foundation coordinates financial and non-financial support to encourage the growth and development of the EOS Network. We’re harnessing the power of decentralization to chart a coordinated future for the EOS Network as a force for positive global change.

// EOS Network Foundation //
https://eosn.foundation
https://twitter.com/EosNFoundation
https://www.linkedin.com/company/eos-network-foundation
https://t.me/EOSNetworkFoundation
https://discord.gg/eos-network
info@eosn.foundation
#EOS #EOSIO #ENF

A

All right welcome everyone. It is july 6th. This is the 22nd week meeting as always we'll just quickly recap. uh Last week it was a quick call. Last week we just checked in on the status of um mandel, upgrade um looked in on platform. Support checked in on all that stuff. We had seven out of 15 bps had signed the m sig for the kailan upgrade at that time, um but really was yeah pretty uneventful. Last week I understand. Maybe we can just hop right into um development updates ted.

A

Maybe if you want to share what came out of yesterday's call so.

B

We um decided we will launch uh 3.1 rc2 today um our egg had some work. He needed to do on that. I'm not sure it's completed it's like six in the morning for him, so he's not on the call but I'll check in with kevin hefner. He may know because he was uh tasked with tagging it um we have. We are waiting for some tests to complete in the trust, evm uh and, and so uh matthias uh is, is working on a bunch of tests.

B

If those tests all pass and give them a higher confidence level, then they'll move to enable the host functions. If the tests fail, then we'll probably give them a couple of weeks and we'll come out with probably something like an rc3 or a patch or whatever that will fix those, but we wanted to get rc2 out now, today-ish, um based on the ship stuff that was in there and some of the uh other um fixes that had gone out already, uh some some p2p networking stuff.

B

um So those are now available and can be tested uh once once the you know the build is tagged and everything and released today, and uh maybe I can ask kevin heffner: did you hear anything from our egg and and did that tagging happen et cetera?

B

Oh, it hasn't happened. Yet I haven't heard anything yeah. It will likely happen by end of day today, so we will then release it and people like matthew darwin can then go ahead and bring it up and start to to work on it. I think there was some question on the call yesterday as to how many of the jungle four people were actually running, the rc one code.

B

Do we have any uh knowledge of that? Is there a good way to check that? How many are running the rc? One looks like a whole bunch, so.

A

We got yeah, we got all these all these folks here.

C

D

C

D

On that list, my apis, yeah, you got to get your your yeah jason.

D

C

D

Are all over the place, so don't get me started that.

C

Damn tool does.

A

E

A

So this is not necessarily an exhaustive list. It does require the bp json um to be uh in a certain state to be able to show up here well.

C

That's the api, that's responding even at that! So but yeah I mean it's a good bearing on it, but yeah we're one extra on that list, but yeah and.

A

I think there may be even others if I'm not mistaken, uh eo sul genie. I think she mentioned to me that um they are upgraded but they're not showing up here either. So that's um something yeah. If you're, if you, if you know you're, upgraded but you're, not on this list, it means there's something wrong with your bp json or maybe looking into.

C

It's on the list, all right.

A

um Okay, was there anything else, ted.

B

um What we can do is we'll post uh kevin. If, if I could ask you after you tag it, could you post in the mandel release the group that uh daniel typically, you know post the meeting notes and stuff in? Would you be able, once you know, it's been tagged and ready to into to post into that telegram group kevin sure. Thank you.

A

Great and yeah, and then I can spread the word to the jungle, um the the feedback I got from matthew. I know he was on the call too he's he mentions. You know, given the the testing still required, it's uh likely a no-go for today for activation of anything. uh We could check back. I guess how the testing goes next week right. um Is that your is that your.

B

That's exactly we are not turning on any of the uh host features etc, because the testing isn't completed, but you can still test the you know: p2p code and the shift code and stuff like that. So there's features in there worth you know exercising um instead of waiting and then having it just a big. You know impulse function of features. We can get some out now and if the testing is good, then we can take the code as it exists and just enable those features.

B

And if the testing is bad, then we'll have to go with it and let them fix the bugs and go with another release or a patch.

A

Okay, uh so he said today's is when this will be released right, that.

B

Was the goal by alright? He had some stuff he needed to split apart and submit, and he thought he would have that done by today.

F

Yeah and I went ahead and approved that uh just a few minutes ago.

B

We should be good, so it's done, you've approved it. So it looks like now kevin it's up to you to kind of tag and build, and all that stuff.

F

uh Still gotta merge, I didn't merge them yet, but yeah we wouldn't do that. Would you will you merge them bucket, or did you do something? Yeah yeah? No, I'm good! Thank you.

A

Great, so we'll shoot for I guess next week again for another, go no go! um I on the sort on that talk. I guess um I don't know if you have, or anyone else has heard um michael from how things are going with hyperion in terms of support is hyperion working with mandel.

C

I have talked to igor who said he was going to start looking into it. I personally tried firing up. He said: hey give it a shot uh it it connected to mandel 31 rc1, but it had the unable to get head block whatever that ship quirk is that they had to fix in rc2 matthew, said yeah, it's probably that same one. So that's why we're waiting on rc2 then can spin up uh or get ship going, and then hopefully we can see if it'll index into hyperion, but rc1 no.

A

Okay, so uh do you think we can make that a priority for by next week? You know so rc2 being out today if we could test hyperion with rc2, that is going to be probably a key decision factor for our go no go next week, I can say, for sure diffuse won't be, which is there's quite a bit of work that we need to do to get defuse working with mandel.

A

So when, when we activate the features, diffuse will not be we'll stop running on jungle. um So if hyperion is also not working, we're limited in history, solutions and block producers are block. Explorers will not work so we'll have to decide hey figure out. Do we have a viable history solution and if not, are we okay living with no block explorer?

A

So that's I guess we can. We can park that conversation for next week, but maybe in the meantime, let's keep in mind that we need to uh validate that hyperion works are.

D

You gonna put it down like just document it there on your list of.

A

Yes, yes, so um go no go.

A

Let's have bps asked to upgrade and test rc 2 this week um need to validate. We have a working history solution.

A

D

Next week,.

D

So makuu your um your ship. Now it's running rc1 right.

C

Yeah, okay, yeah and it connected to it, but it said unable to read blocks and that's what matthew said the rc2 ship fix is supposed to address. I believe so. um Good news is, I don't think it was really mad about it. It just couldn't sink. Couldn't keep up.

C

I think if somebody head block um yeah okay, but I also talked to igor who said he was gonna, try to take a look at it or whatever, but he's probably busy, but um it is at least on his radar, but he was like yeah give it a shot, it might work and um it didn't blow up as much as like 2.1. You know the 2.1 parts are hated.

G

Yeah 1.8 it was, it was happier.

C

With it but um yeah it just, it was stuck, but that's rc2 and yeah whenever it drops uh I'm assuming yeah, there's no log changes or anything like that. So I should just be able to drop the binary in place and and hopefully start testing pretty quickly.

A

And uh and then once when it's I think we should be pretty quick to upgrade everyone. I think I had heard there were some challenges with stuff for some people getting upgraded to mandela, especially for those on ubuntu 18.

C

If you're on 18, it was miserable- and I just I gave up- I think ross gave up on it. So like say I it took me five minutes five minutes when I gave up yeah yeah, it took me a lot longer than that, but also break a lot of more stuff, but anyway.

G

C

Short, I mean if we want to say we support 18, there's a lot of work to be done. Even I needed ross's. You know here all these prereqs um for 20 to work, but my concern is that you need 20, that's going to be a pretty big ripple effect of yeah. I mean you're about to have to deploy new servers or go through os upgrades. You know I mean it's you're going to trigger a infrastructure upgrade which is good.

C

I mean we've been stuck on 18 forever, but it is brutal to even think about it on 18. at least building it. So I.

D

Don't think you don't see it as a problem honestly, I don't see it as a problem.

C

Yeah, if he was 18.

D

C

D

Yeah I mean you're going to be spinning at new infra anyway, so just start building 20. I I wasn't. I was going to start upgrading some nodes.

C

But yeah I mean.

D

You know what I mean I was replacing them.

A

Yeah, if I I think I've heard, though, that there's the documentation was just updated, also this week to give some more guidance. If I'm not maybe someone on the from the develop development team can add some color to that that documentation will.

G

A

With instructions to guide upgrading from 18 more so it.

G

Should be possible.

A

To upgrade from from, if you follow the the specific steps which are now documented.

A

And that's that's just in the in the mandela read me is that right.

G

Matt and you I, I ended up fixing a number of problems with the 1804 build uh over the last couple weeks. uh Great expectation is that expectation is that that should be uh clear sailing there. So you want me to try.

C

It on 18, I I'll bang my head against it. A little more broth, probably won't I'll, try it. I.

A

Think I think it's worth a test. Yeah yeah.

D

If 18 is available well, then I'm going to use it. It's 18 goes until 2028 or something or 27.

B

F

Right, I think, I think, for.

D

A support curve.

F

Is that it should, it should build and work fine on 18, 20 and 22., so it's worth giving it a shot. It's been updated and.

C

Hopefully addressed so perfect that was.

G

C

I wasn't even going to try anymore.

G

Yeah, I've personally run through all three of those, uh so I I'm pretty confident it's going to be fine for you, we'll.

C

See if it's eos, usa proof, let's put it to the test.

A

And again, if you've already upgraded to rc1, it's going to be exactly the same for rc2, so it should be, I think, quicker um without getting getting folks upgraded to this next release.

F

I mean you just got a replay, there's one nice thing there that we also should have uh binary packages to support that too. So, if you don't want to build it, um there's a good chance that you don't have to. But if you do want to build it, then by all means please test it out.

F

G

Bps build for rc1 we yeah for rc1. We only had binaries for 2004, but for rc2. The intention is to have it for all three.

D

Great just looking, I was just looking at the releases, so it ends support 18 ends um standard support next year. Right, I don't know what I don't know, what the difference between standard support and their life ender life is 20 28, but uh yeah, so it'll keep getting security updates until 2028. yeah. So you put you probably you're, pretty good with it for a while. Well.

B

They officially support it for five years, but they throw security stuff. On I mean we we should be. We don't want to keep supporting more than three packages. So 2022 is now the current version. We should say 2020 and people should be moving off of 2018., because by 2024 we're gonna, we're gonna, deprecate it and say it's gone because we'll then we'll we'll do 2020, 2022 and 2024.

B

2022 is actually lts. Yes, it's alright.

D

I tried to run it the other day for some other stuff and there was all sorts of brokenness. I was like.

C

Years ago,.

D

Yeah you're always good to just sort of hang back a couple of years.

B

Usually, if you wait till october, it's pretty solid, they give you the first point: release yeah.

A

Cool, so is that is that worth a bullet here, the we've improved the support for ubuntu 18. Yes,.

G

A

Point something I'd say: yes, because it you should try it again. If.

C

You tried it before.

B

A

All right and that's just specifically an rc2 or across the board.

B

I think you have to pick up rc2 matt. What's the answer on that? I.

G

I well yeah, I mean none of those changes were they're, not back released. Yeah yeah, I mean they're not back released into rc1. um I guess the instructions, uh some of the instructions would probably apply to rc1, just that the actual just like the instructions but um like the scripts would wouldn't it wouldn't have been backboarded. So, okay, let's tell them to use rc2 move forward.

A

Okay, uh anything else to add on, I see, there's a question here from kevin he's interested in feedback. um If phil pg am I saying that right is something that should be updated and maintained.

C

I don't know anybody actively running it. I know some people have talked about it over the years, but the problem is a lot of those just get implemented by you know: people like liquid studios, they just build their own monstrosity and then it's like oh yeah, I'm using that you know. um So you don't. I don't know of anybody actively, but I do know I've seen developers reference it, whether they implement it or not.

C

I don't know you'd have to ask that survey. Maybe.

A

Yeah is that something we want to survey on this is: is this regarding prioritization for mandel 3.2.

E

uh Yeah it'd definitely be a 3.2 thing, uh it's just something that came up and uh ted wanted me to remember to ask in this group and so uh yeah I mean, I think it's just whether you know we should expend resources to get it up to date and working. I believe it. It never uh worked for 2.0 of esio.

E

It was updated to work with 2.1 of esio, but it would take some work. I think, to get working with with manual.

C

Oh, if it doesn't work with two os and yeah, I doubt a lot of people are heavily using it. What you may see a stan, maybe because I know he's done a lot of those integrations. I know he just built a whole system for tracking traces and something so it may not be as needed, or there may be some sort of a if you're still using it. You need to take old, bessie out back and maybe use one of the other technologies, but that'd probably be a stand.

C

Question I don't know what exactly all went into phil pg. I know he's tracking traces.

A

Okay, we can send out the horn anyone else on. This call don't have any.

H

Just as a side note, our our history solution will also support postgres, mysql and tidb.

H

So if someone wants some form of relational database with full history, we will probably support it.

C

It's gonna be open source.

H

C

H

Yeah, so that's what I say.

C

Is there's probably gonna be some community tools that I know some of those gaps are being filled. So I'm not the person to make that decision, but I would potentially say probably not especially if it was pretty broken to begin with.

B

I think it, but it did work with 2.1 right kevin so for people that were using it on api nodes and stuff like that, it was working.

B

um That's why we were kind of wondering if we could get a survey or something to see if people care about it, because we can have oci, get it back working again. um What is the other solution? uh C c: mad h.

I

H

I

We are basically.

H

We are basically working on on a diffuse light.

G

H

Because the fuse is so so expensive to host and right it comes with so high maintenance cost uh yeah. We are building a light version of the diffuse and it will support different uh storage backgrounds.

B

And that's liquid, I o! So if we just went out to liquidio, we can see that and people could download that etc. No.

H

It's in our in our github, I mean it's not production really, but we have.

G

H

Prototype and constantly constantly adding stuff to it um and there will probably be a grant application uh in the in the eos network foundation, grant applications repository.

B

Yeah, I was about to say, like I, haven't, seen your grant application, yet I'm the guy that is the first wave of processing I'm working on it. I'm working on go ahead. Yeah contact me if you need any help.

C

He should play his game. It's pretty cool too.

H

Yeah, there's there's a lot that we're working on it's yeah.

A

You gotta eat your own dog food.

C

A

C

A

It right I mean I remember that conversation.

C

Talking about how diffuse was the only thing that could keep up, and even still it was heavy.

H

Yeah yeah yeah, the the I mean, the the reason we are building diffuse light is basically we have been waiting three four years for someone to to either build a diffuse that is um um so low in in maintenance and and hosting costs that everyone can host it or our ecosystem to thrive. So much that everyone has the the money to host it. But neither is the case so uh yeah and there are a few crucial features and diffuse uh that we think are valuable to the community. So.

A

From from your perspective, corvin, uh I don't know how you have you done a good comparison between, like just the fire hose part of defuse and a full defuse in whether the fire hose alone. The fire hose, is way more lightweight to run sure.

G

A

Doesn't have all the features of defuse um and the interesting thing about fires is: that's what's evolved into what's turning into the graph, you know when the stream fast team got basically acquired by or uh granted.

H

By the by the grad foundation,.

A

H

The fire hose is absolutely nice, but it you also need some form of back-end service to to consume and process fire hose blocks, um yeah to to to post-process them, to put them into some form of storage, backer and etc.

H

That all comes with latency. You still need to to host the files, application um and yeah for different reasons. We decided to to to build a cold.

E

Solution cool is that built on ship or is that using the deep mine, water.

H

It's using deep mind: do you mind one of the biggest the biggest benefits in our eyes is that you have full deltas, full deltas and mapped deltas. One of the big problems with ship is that deltas and action traces come separate and there are. There are applications like like the atomic assets api.

H

I think it's called usao contract api.

G

H

Some service that you can run on your server, which tries to map action, traces and deltas together on based on some form of estimation,.

I

Magic yeah.

H

Yeah yeah and I think for gaming, and also for for d5. It's crucial to have that all in one data package and to to have full letters and not only like the the end are deltas that the chip provides.

H

A

Okay, um so on that field, phil pg topic randall- is that something you can help with another survey.

A

And if you can't speak now, we can catch up offline.

B

Or maybe text into the chat.

A

Or text in the chat.

E

Curious, what's the what's the overhead of running the deepmind logger, do you find that that overloads a node or it's not that much additional on just a normal node, that's running.

H

Just just running the deepmind logger doesn't doesn't make make a huge difference, at least so far. We we haven't measured any any any real overhead, any any additional measurable delays. It replays with the same same performance as without.

C

Actually, testing stress testing how many services can can rip data out of a ship node. So I'm curious, have you loaded it up? That ship seems pretty actually efficient at handling a monstrosity of eight or nine servers streaming wax. So I'm curious if you really tried from that aspect, seeing how it handles it only because I've been recently hammering ship to stress it.

H

I think the maximum I tried with ship was four or five notes consuming the the web socket and stuff.

C

It can handle more just so you know, but I mean obviously there's gonna be a limit, but I'm curious if you've done some stress testing. If I was surprised at how well it just handled extra network load, you know it really didn't hammer it that much more it didn't seem. So I was curious if loading up deep mind consumers made it happen.

H

There's only one deepmind consumer, I mean it's writing to std out and you basically read from std out and then pass the binary data into in our case.

H

C-Sharp objects in diffuse, it's passing them to go objects, and then you, you do all the the post-processing, but there's only one one consumer of a demand: log yeah in theory. You could run multiple but.

C

Well, that's what I've been doing like.

H

Five atomic.

C

Servers attached to the same ship, node and three hyperion servers.

H

To the std out when transactions are processed and blocks are produced, you can't say hey deep mind, give me block 500, you only get what's currently processed.

C

I got this, it's a whole different mentality.

A

Let's, uh let's take the rest of that conversation offline, but that's good to hear that we've got some options on the horizon for everyone. um Speaking of the survey looks like randall is listening. He can't he can't speak, but he did confirm he can uh help us without with by sending out a survey on that question here.

A

So that's great. Thank you. Randall uh speaking of surveys that randall and the eos support team have helped. With um on the platform support topic, we had sent out a a survey with additional feedback um and we have some responses here. I can share again: randall can't speak now, so I'll I'll just quickly share. I don't think there's value in reading through these most of these are from people who haven't actually tested dune, um most people just saying yeah. It looks good from what I can see. um One person.

A

Is this the one um some concern here again, not someone who's actually tried dune um just says doc. Machine.

B

A

B

Which is that's yes,.

A

Okay, so there we go that I'll I'll share this file in the I don't have it yet, but I'll add it to the um this platform support here in case, anyone on the dune team wants to uh wants to read it but, like I said, yeah, not not a whole lot there. Besides a little bit of.

B

Is there, can you just scroll down a little bit um just see what last guy said.

A

B

A

This one is, is again, I guess a misunderstanding of.

B

Hears that dune is not helpful at this point, which is the exact opposite of what we just did.

A

Right so I guess I think it sounds like he was listening in on that call. When we had artem on and and the feedback was that, as you know, it was seemed more targeted to node operators than smart contractors. I think all that has been resolved and again, that was more of, I think, a documentation, clarification thing.

A

um So more of a misunderstanding of of dune's target audience.

B

I almost think we need to do either a better job of marketing it or you know getting out to people and explaining it because they're hanging on to old information of you know docker and stuff- and you know dune is not the same thing as the old docker process. So we've specifically made changes to support m1 and it works a lot faster now. But you have to try it to figure that out.

A

Right right, okay, cool! So again, I will share this. um Do do what you like with it and um a little bit, maybe we'll at some point will maybe when we have.

A

um If, if we do do that marketing, we can kind of re, follow back up with these folks and see if see if they change their their feedback. Although again most of these folks didn't really have much negative to say besides this one or maybe this one, um okay, anything else to add on the platform support side of things. I think, last time we checked still some documentation in progress.

F

So it's not really documentation it's just um effectively using the binary packages that we're going to produce for the for the rc releases so friday we're going to kind of do the the v1 of doom there and then that one will actually attach to the rc2 releases and then thursday, we're going to have a cdt release, so that will be sort of the first version that finally gets to use these packages and be able to do the contract debugging and all that kind of stuff. Yeah. That's probably worth.

B

Noting that there's a rollout schedule that thursday will be cdt and contracts and friday will be due and today is the main mandel release. Okay,.

A

So that's not all platform. This is more as software development side of things right, um cdt. Let's go over that again. Okay,.

B

A

B

B

And dune on friday.

A

And that's like the dune with support for rc2. Is that right? That's the the major change.

F

Yeah he's the full dune at that point, the stuff that artem was wanting all that will be there at that point.

B

Yeah great, noting that it includes the stuff artem had wanted for the osao tester.

A

Testing and debugging was it.

B

Well, it's it's a specific type of tester called called eosio colon, colon tester, and it's it's. uh It was not really meant for end users to use, but some people found it and used it and artem was one of them and it wasn't supported in dune in the first release, but it will be in the rc2 release.

B

B

You have eosin and you mean ice. Thank.

F

A

And support for rc2 will be released friday.

B

A

Okay, any other notes to add that covers it all right.

A

A

um Okay, so we covered software development covered platform, support testing on jungle. Four we looked at, we were gonna, be kind of briefly touched on it. We got the more more bps are now upgraded, updated. We've got better support for ubuntu 18..

A

um Anything else to note regarding testing. Has anyone come across anything any other than any of the folks who've recently upgraded.

A

Otherwise, just um again we'll we'll we'll promote the once rc2 is ready to go, uh we'll promote it in the jungle test, get folks to upgrade and and, uh like I said, request especially testing for history solutions.

A

um Then we got oh yeah, so we're making progress on our msig on kylin. It looks like we're up to 13 out of 15.. I think that's probably still the case so just need a couple more vps to sign this and uh and then we're rolling on kylan.

A

So we got you know. Aloha can push us over with just on their own essay asia. Canon sg so we'll continue following up and checking back in next week on how that's this is going.

A

uh Are there any other topics.

A

Well, it's a good chat today, uh we'll, uh like I said, see you all online in the jungle, chat and uh and happy testing thanks, guys, excellent, all right cheers everyone take care.

E