diff --git a/wg-provider-audit/001-2024-02-21.md b/wg-provider-audit/001-2024-02-21.md new file mode 100644 index 00000000..7bf6e9a0 --- /dev/null +++ b/wg-provider-audit/001-2024-02-21.md @@ -0,0 +1,344 @@ +# Akash Network - Akash Network Provider Attributes Working Group-Meeting #1 +## Agenda +- Welcome and Introduction +- Overview of Working Group's Purpose +- Review of p-i-3's Progress and Next Steps +- Discussion on Meeting Cadence and Timelines +- Any Other Business +## Meeting Details +- Date: Wednesday, February 21, 2024 +- Time: 10:00 AM PT (Pacific Time) +- Recording +- Transcript + +## Participants +- Alani Kuye +- Andrey Arapov +- Anil Murty +- B S +- Benjamin B +- Deathless +- Luna +- axime Beauchamp +- Maxime Cyr +- Rodrigo Rochin +- rUqeRT +- Scott Carruthers +- Tyler Wright + +## Meeting Notes +#### Tyler Wright: +- Tyler Wright welcomed everyone to the meeting. +- The working group emerged from Sig providers, focusing on tracking the p-i-3's work on building an open-source automated auditor on the network. +- A proposal was put up on-chain at the end of 2023, which passed. +- Tyler shared the proposal in the chat. +- The group aims to monitor the p-i-3's work and address any questions or concerns. +- Dedicated sessions will be held for discussing project-related matters. +- Cloudmos team's contributions were acknowledged. +- Currently, the auditing process is managed by Andre, with a template available in the community repo. +- The working group intends to improve the auditing process over time. +- The focus of the inaugural meeting is to check in with the p-i-3 team and establish meeting cadence and timelines. +- Benjamin was handed over to provide updates on the p-i-3's progress and next steps. + +### Benjamin Update (p-i-3 team) +- Learning GoLang for contributions, along with Sebastian. +- Planning to structure the project by creating prototypes first, then integrating into Cloudmos. +- Conducted small testing for benchmarking system capabilities. +- Challenges with accessing working temperature data. +- Seeking options for outsourcing KYC updates and verifying companies' information globally. +- Requesting ideas or connections for improving verification methods. +#### Questions for Discussion: +- Are there better options available for outsourcing KYC updates? +- Ideas on how to verify companies' information globally? +- Suggestions for improving verification methods? +- Open floor for discussion and suggestions regarding the questions raised by Benjamin. +### Anil's Updates +- Anil Murty suggests expanding benchmarking tests to include parameters available through Nvidia SMI command, such as fan speed and temperature. +- Anil Murty mentions previous use of KYC services by the labs team, potentially offering options or recommendations for other providers. +- Benjamin expresses interest in KYC solutions with minimal involvement of third parties. +- Maxime Beauchamp discusses benchmarking methods involving specific workloads or Docker containers on providers, potentially integrating benchmarking into the provider service. +- Benjamin and Maxime discuss the feasibility and benefits of integrating benchmarking into the provider service natively. +- Benjamin acknowledges the potential cost-saving benefits of utilizing the discovery service for benchmarking. +### Scott's Updates +- Scott Carruthers discussed the upcoming release of a functionality called feature Discovery within Provider Services. +- Feature Discovery is accessible via grpc and provides visibility into individual providers and across the network. +- Full GPU Discovery is a key aspect, offering information on GPU types, allocation, and availability. +- Worker nodes with multiple GPU models will provide detailed information on allocated GPUs. +- The functionality will enable analysis across the network, indicating total GPUs and allocation status. +- Rollout to strategic providers like Shimpa and Rodrigo is underway. +- Benjamin B acknowledged the inclusion of CPU information and potential for future additions based on demand. +- Information is accessible via direct grpc queries or the Cloudmos API, with a provided grpc query link for exploration. +- Participants are encouraged to explore the provided link for firsthand experience with the feature. +### Potential falsification of resource information by providers. +- Luna raised concerns about ensuring providers do not falsely advertise their resources. +- Scott Carruthers emphasized that feature Discovery aims to address this issue by providing constant, real-time updates on resources through a grpc stream. +- Feature Discovery operates at a PCI bus level, not relying on third-party binaries. +- Luna inquired about potential vulnerabilities at lower levels beyond the PCI interface. +- Scott Carruthers acknowledged the possibility but stated that, currently, there are no known limitations or concerns regarding feature Discovery's effectiveness. +- Benjamin B suggested that, despite safeguards, there may still be ways to falsify information, but addressing such issues would require time, incentives, and further exploration. + +### More Questions (Benjamin Update p-i-3 team ) +- Benjamin B raised a question regarding the accessibility of global databases for certain countries, noting potential challenges in obtaining data. +- Alani Kuye inquired about the process for provider checks, audits, and incentives, particularly regarding participation in incentive programs and the audit process. +- Tyler Wright redirected the discussion, suggesting that questions regarding provider incentives and audits should be addressed at the Sig providers level. +- Tyler clarified that a specific working group is primarily focused on developing automated auditing tools, while manual audits are conducted as part of the provider incentive pilot program. +- Further discussion on provider incentives and audits is scheduled for the Sig providers meeting next week. + +### Functionality and usage of the "provider TX audit attributes" command. +- Benjamin B resumed the presentation, mentioning the possibility of reading data via API instead of relying heavily on CLI commands. +- Benjamin posed a question regarding the ease of using the "provider TX audit attributes" command and sought clarification on its functionality. +- Andrey Arapov explained the process of broadcasting transactions on the blockchain to sign attributes and monitor changes, emphasizing the importance of ensuring attributes are signed. +- Andrey clarified that while it's theoretically possible to broadcast specific attribute changes, the tool typically broadcasts all signed attributes. +- Benjamin discussed potential future uses of multiple wallets for signing purposes and expressed gratitude for Andrey's insights. +- Andrey offered to provide further information and requested to be pinged on Discord for assistance. + +### Project Deliverables and Timeline +- Benjamin B expressed gratitude for the shared link and concluded the presentation, addressing any remaining questions. +- Tyler Wright raised the topic of project deliverables and suggested establishing a cadence for progress updates and testing phases. +- Tyler proposed a monthly cadence for the first three months, transitioning to a weekly cadence thereafter to meet the proposed timeline. +- Benjamin agreed with the proposed timeline, noting that most tasks will likely be completed simultaneously rather than staggered. +- Sebastian confirmed agreement with the proposed meeting schedule and timing. +- The team agreed to continue monthly meetings, with potential transition to weekly meetings in the future. Meetings will continue on Wednesdays at 7 pm. + +### Final words +- Tyler Wright proposed creating a plan for the first deliverable to be discussed in the next meeting, along with a short testing plan, to ensure tracking of progress for those unable to attend meetings. +- Benjamin B expressed a preference for online discussions over using images for benchmarking, inviting input from team members on alternative approaches. +- Benjamin B thanked participants for their contributions and assistance in answering questions throughout the meeting. +- Tyler Wright confirmed plans to create a specific repository for the working group on provider audits, containing meeting notes and relevant documentation. +- Tyler extended gratitude to all participants and encouraged continued involvement in the Discord community. +## Action Items +- Benjamin B to further investigate the possibility of a solution for KYC with minimal third-party involvement. +- Maxime Beauchamp to explore the feasibility of integrating benchmarking into the provider service and report back with findings. +- Andrey Arapov to rovide additional information and guidance on the broadcasting of transactions for attribute signing. +- Tyler Wright to create a specific repository for the working group on provider audits, containing meeting notes and relevant documentation. + +# **Transcript** + +_This editable transcript was computer generated and might contain errors. People can also change the text after it was created._ + +Tyler Wright: All right, welcome everybody to the inaugural session for the working group provider audits. + +Tyler Wright: This specific working group, which has come out of Sig providers is designed with the intent of tracking the work that the pi 3 team is doing related to building an open source automated auditor on the network. they put a boat up on chain and that boat passed gonna drop it in the chat for folks. And again, this happened at the end of 2023 and they've been working this part to get up and running have attended a number of meetings to again get themselves up and running but we are going to have a working group related to just again monitoring the work if the high three team has any specific questions that they have obviously they know that they can reach out on Discord or come to six support or other very sigs. But again, we'll have a dedicated session. + +Tyler Wright: where we can go over anything related to the funds anything related to the project that the pi 3 team is going to be working on. I know that part of most of that project is around again an automated auditor solution, but I know there's also some most miscellaneous contribution being made to Cloud though so much appreciate the cloud most team in being here. I'm not obviously right now the process for auditing is managed by Andre and folks there's a template in the community repo for folks that set up providers and want those providers to be audited by Andre. But again, this should improve that process over time. For this inaugural meeting. I know that we just want to check in with + +Tyler Wright: the high three team and see where they are at this point. I know that they might have some questions and then we can figure out this meeting today or at least talk about it right now or talk about it today and then finalize between meetings just a Cadence for how often this group will meet and just a general timeline. I think the pi 3 team will share with us on when deliverables should be expected. So without further Ado, I'll hand it over to Benjamin who I think has something to share and we'll talk at a high level about where the pi 3 team is and next steps. + +Benjamin B: Thank you I think this introduction was Enough, so I'm not going to talk more of that. I will instead share my screen and I will get started. + +Benjamin B: So before we show our actual updates I want to say that we have some points here that we have done before and outside of anything. You'll see. for example, because a contributions are mostly preferred to have in either golang or in the node.js. And Sebastian has been working on learning golang. I will also be doing this for my contributions, but I'm not started on that. Not yet. then we have a lot of internal planning on how we want to structure everything and to speak briefly on that. We have made the choice to First create some prototypes. + +Benjamin B: And then create the actual systems. So we'll first get it up and running using our own systems and then we'll try to after that integrate it to cloudmus will integrate. To Cloud almost again and maybe work with the greater team to get started with. Some of our stuff as well. So for the benchmarking system updates. We have done some very small testing just to see what's possible or This is not a lot of to show for here. We've been able to get for example the capabilities. For this thank you. So what instructions you can give to it and it will work fine. + +Benjamin B: the product and vendor names so for example and vidy and + +00:05:00 + +Benjamin B: 1070 if we were speaking about gpus. And for CPUs, it will be something in the land. Or something very specific. But let's just say I 700 by 7,700. then we have + +Benjamin B: more things as well We also have the figure out how to read the memory size for the entirety of the rent of the working load. We also want to try figure out if we can do this for just the party renting but that's irrelevant to the fact checking the resources that's more of You get what you're paying for maybe? And then we also try to get the working temperature out and at least in the configuration. I was unable to do it. because of access and after searching a bit this kind of made sense. We want to do. more here And we're not going to start doing the action benchmarking it. for the kyc updates + +Benjamin B: We have Some options here. but instead of going in depth over what's this is and I just want to go over to the questions instead. That I have here. Is there anyone here that knows if there are better options available Even better if this could be outsourced to someone in the community. maybe if you know someone who deals with this sort of stuff, it's a very specific ask but you never know. + +Benjamin B: And I'm also wondering if anyone has some ideas on how to verify companies in another way. I know that in Sweden, we're able to have Some contact information websites Etc. That are linked to your company. But I was trying to find some Global source for this and I was only able to find it. I was thinking about maybe you can connects to The Domain in some way, but you already do that when you least from the provide around even And look up the domain. You don't know if someone has lied about the information. So I'm not sure how to continue There. + +Benjamin B: If you have any ideas, please raise your hand now on either this questions. + +Benjamin B: Feel free to go ahead. + +Anil Murty: Hey Thanks for putting all this Here for us review one question and one comment. So on the benchmarking Test that you're talking about. I know you mentioned gpus, but could you extend that if possible include a bunch of parameters that are available through the Nvidia SMI command? So it will give you things like The fan speed the temperature out of those things from the Nvidia semi command. They might be a little bit of passing that you have to do, but it's pretty extensive what you can get from it. And then the comment I had was as far as kyc as a service goes. we meaning the labs team have used kyc in the past for various tests. So we might have some options there, but + +Anil Murty: In general are you just looking for recommend suggestions for other kyc as a service providers? + +Benjamin B: Yes, either that or if you can offer some sort of the solution where the idea of the thing would be if providers could give their ATT to someone and that's someone or not company or whatever could do the kyc. Of course, it's possible to do everything with another third party, but the less part is involved the better. It's rather one that wants to speak more of something or should I go to the GPU question? If so, just please free to go ahead now. + +Tyler Wright: I saw that Max and Scott Heather hand up Matt. Did you have something specific about the land about kyc or was it about just general Benjamin benchmarking? + +Maxime Beauchamp: It was about the benchmarking. So if someone else wants to talk about the kyc I can wait. + +00:10:00 + +Benjamin B: No, no, go ahead with benchmarking. That was the first part of the question. + +Maxime Beauchamp: And I was talking with archer this morning about that topic specifically Because right now I assume that the way that you want to do the benchmarking is to run like a specific workload. Or a Docker container on each of the providers… + +Tyler Wright: Good. + +Maxime Beauchamp: if I'm not mistaken to get that information. + +Benjamin B: Yeah, we want to both do something where we can read off information. And then we also want to run some sort of image that needs specific work and if that retires the attributes are actually real or not. So perhaps if you have a certain CPU Euro offering maybe we can even go as for us. To check or the capabilities the same by actually testing the capabilities. I don't think we're going that far but more like running some workload and seeing how fast an extent and compiling that's data. + +Maxime Beauchamp: okay, and so yeah, I was thinking further about this and I think ideally because right now Scott will probably talk about this but the team has been working on releasing the provider Discovery service which kind of does that exposes more information about the GPU right now, but I'm pretty sure it can be extended to and more information related to the CPU memory and stuff. and when I was talking with archer this morning is I think the best way to do benchmarking. Would be to integrate it natively in the provider service. because for example if the providers don't have any space available to run the benchmarking. + +Maxime Beauchamp: Workload, you can't Benchmark that provider if it has a white list or something like this. There's a lot of constraints. to run a workload on the provider. So I think the best way to do this and if you mentioned that you want to learn golang, I think the best option for benchmarking would be to introduce it in the provider service natively. + +Benjamin B: Yeah, I think that makes a lot of sense. I still think it could be worthwhile to verify it in extent. But maybe the benchmarks will speak for themselves. I know that Luna wants to talk a little about how the attributes work. But if you would solve this through the discovery service instead, I'm guessing a lot would be saved in fees. So this could be an objectively better solution. I also see that Luna erased. their hand also if you want you can speak more on this Banner. + +Tyler Wright: I think Scott had his hand up. + +Luna: I think Scott yeah. + +Tyler Wright: I think Scott had his hand up and then Luna can go right after. + +Benjamin B: Yeah. + +Scott Carruthers: Yes, Ben. I do you mind if I just steal the screen sharing for one second? + +Benjamin B: Yeah, Let me. Close it. + +Scott Carruthers: Are you able to see my screen? okay, so to part of Max's point,… + +Tyler Wright: Yes. + +Scott Carruthers: so I just want to make sure that there's no redundancy and also that your application and your efforts can draw on from any existing information, so were about to release a functionality with them Provider Services called feature Discovery. And if you look at the terminal this is exposed via grpc. So this is a provider in sandbox that is currently running the feature Discovery service. So if I do a grpc query of that we can see that as far as + +Scott Carruthers: Let me get up to node 2. + +Scott Carruthers: So you can see that in part of feature sc We're doing full GPU Discovery. So this is going to give us full visibility into individual providers. And of course across the Expanse of the network as well. So this is just a provider with a single Nvidia Tesla T4. But if it had + +00:15:00 + +Scott Carruthers: if we have worker knows that have several different models, we'll get information on the type of gpus in the wall. So get allocated and elocatable information so we can see that all and in this section. So we're doing GPU Discovery and then we're also able to tell on those worker notes. How many are El capable essentially total gpus and also So this is a provider that had a total gpus and had for concealing we would say allocated for and then as I said, this will not only give us visibility for your perspective the most important thing would be on a provider basis. But then also this is going to give us the ability to be able to do analysis across the network and be able to say, the Akash Network currently has 100 a100 and 75 of those are currently allocated. So this is something that we're actually currently rolling out to a number of strategic providers, like shimpa and Rodrigo it's about to go through + +Scott Carruthers: Great on his provider. So we just want to do a little bit more sending a checking and a controlled Manner and then we're going to start doing a network where I provider upgrade. So the end result of that is feature Discovery and what I'm showing you should be available on the network within the next several days. gpus while the most important driver for feature Discovery, we're also doing things as a GPU and memory level as well. So again, I just want to make sure that your cognizant of this or this might be some things that you can draw into your application instead of doing some of the Discovery yourself, I just want to make sure that you are aware that this feature is imminent. + +Benjamin B: Yeah, thank I see that even some CPU information. It's available here. And I'm guessing that… + +Scott Carruthers: Yes. + +Benjamin B: because you're able to get the vendor you're probably able to get everything else to that in some way at least so… + +Scott Carruthers: Yeah. + +Benjamin B: if there's demand for anything more this can be added on to at any time. + +Scott Carruthers: Yeah, okay, and I will share in the chat and this goes for anyone that's kind of interested in saying this as you can see. I'm just collecting. So another important point in this is this information is all being consumed by Cloud most into their indexer. So you could get this information through direct grpc queries to those individual providers. But soon you also be able to query they Cloud most API to give feature Discovery information. But if you're interested in this information, I'll post this grpc queer Link in this is a provider that's going to be up for some time. So if you want to go in and play around with this at all yourself and see what information is available, you'll have that. + +Benjamin B: Yeah, please feel free. + +Scott Carruthers: Yep, that's all I had. + +Luna: All Thank I think this is a general question and it kind of ties with the feature discovery. And it's really just how are we making sure that providers don't? + +Luna: don't bake or falsely advertise their resources. So for instance, we might be getting GPU information from an binary. I think it's Nvidia semi, right? but I think that could easily be replaced with some malicious or fake binary that just Returns the information that provider once so we can falsely advertise resources that he doesn't have is it something that we are concerned about or not? Really? + +Scott Carruthers: Yeah, you know what? We're absolutely concerned about it and say if anything that was I would say that there's two primary drivers for feature Discovery. One of them being provider incentives being imminent and we want to be able to not only an appointed time but consistently through a grpc stream be able to tell if resources are removed inserted into system. So one of the drivers for feature Discovery is for provider and setas we're going to have a Constance and hack of not what providers are advertising via attributes manually but through this Hardware Discovery process. So yeah, that was certainly one of the big drivers for future discovery. + +Luna: I think it doesn't the feature Discovery software per se also have the limitation that it could be faked. So what exactly is the software querying I think if it uses some system binaries, they could easily be mimic to show information that would be false and does the provider will also relate it or the feature recovery allows all the relay that false information. + +00:20:00 + +Scott Carruthers: Yes, so the feature Discovery is not relying on any third party binary. We're doing Discovery at a PCI bus level. + +Luna: okay, and every made any discovery on + +Luna: possible ways that it could still be fixed or is it another concern at that level? I'm not familiar with the PCI level. + +Scott Carruthers: so I could not somehow be manipulated at a hardware level. I mean it's possible I think for the current. + +Luna: I mean, I mean not at the hardware level but at the lower level than the interface that you are attracted. + +Scott Carruthers: yes, I guess as a short answer would be at this current period of time so obviously if we Encounter any of those kind of concerns over time. We'll certainly address them for the initial purpose of feature Discovery. As I said, the primary goal being that we have a consistent view for provider incentives. So obviously we could always get into the provider ourselves through test appointments and things like that and do further. Station, but for the purposes of what we initially need feature Discovery for and for sending a checks of Provider hardware for provider incentives. I'm not aware of any limitations or concerns + +Benjamin B: Yeah, I'm guessing. in some way it should be possible to fake it regardless because + +Benjamin B: a culture doesn't really run your computer. You're the one with the access to it, but I'm guessing that this will take time to figure out and should also need some incentives to do it. All right. There was one question. That was not fully answered. I think. And I believe it was related to this. + +Benjamin B: and now I know we got a message in the shots that mentioned that their likely sent a global database. but for some countries it's possible to get this data easily for some other countries. It's not as easy. So thank you for those from that a lot of episcopes + +Alani Kuye: Hi guys. I just had a question around incident, Senator checks and Audits and incentives. if they providers are up, I mean first things for us are the providers opted into the incentive program or every provider just gets it that's the most high level question and secondly won't that be predicate on them? + +Alani Kuye: to get incentive then you have to obviously be a part of the audit process, right and the audits. I don't believe they'll be scheduled and they just happen on an ad-hoc basis that way, you don't have common providers. No one deal. It's gonna be and making sure they're ready to pass it and then they go back to the usual antics. + +Tyler Wright: I think that'd be more of a question for Sig providers at the highest level. I know there's operation in place to make sure that anybody that's participating in the provider incentive pilot program is getting audited with regularity and that'll be mostly manual but specific to this working group just push it forward. This specific working group is more mostly focus on the provider auditing automated tool. That's the high three and his team are working on. So if we can save that conversation for six providers, which is happening on schedule next week, that would be terrific. Unless anybody have yeah. + +Alani Kuye: Yeah. Thanks Tyler. + +Benjamin B: Did you have anything else or? it's memo. Okay, let me share my screen and I will continue with. The presentation and then Anil if you can send over more information. At the later point when more about the kyc provided you're using and if it's possible maybe to what I mentioned, that would be great. + +Benjamin B: Okay, I can see it for the auditor and this is the last part of the presentation. + +00:25:00 + +Benjamin B: We have seen that it's possible to read the data with API. This means that we don't have to utilize CLI in the same way that we originally didn't know if we would have to rely on that much. And then for all the thing, I went through the CLI commands and I have a question. Is it as easiest Around the provider TX audit attributes command. But then sharing the shot. Or is it the bombing another way to actually command for it? This sounds super simple question, but it looked more to be related to they actually provider, but maybe this is how we do it. I'm guessing someone here should know how the sign. command Works, especially Andrew + +Benjamin B: there's no need to talk about this here if you don't want to but that's just an open question. I'm hoping to get an answer. I'll ask it in Discord. + +Andrey Arapov: There is just a broadcast the transaction on the blockchain that this certain attributes were signed and trying to address the attributes and if the provider adds new attributes, they have to be resigned and something that I have script for a monitoring from time to time once a week plus minus and if I see some provider changes attributes like this, I'm just like a sharing the transaction again to make sure that the new attributes are signed and if they have removed a lot of stuff I can remove there is a broadcast transaction to remove what's been signed. So to make sure that the older ones aren't signed but not so necessary in some cases, example, for example, if provide your changes better to better three type of storage and then changes back so there is no reason going on sign the previous ones because they can change any time. So the better for us to make sure the sign because we trust him anyway, so + +Benjamin B: Yeah, I understand. and what you're doing here is essentially that all of the current attributes you're signing and then you can again do all of the current attributes. I'm on signing and then signing again. Or do you do with them or… + +Andrey Arapov: Yeah, if it's not necessary to unsign… + +Benjamin B: attribute basis? + +Andrey Arapov: then we know you can just like a broadcast again to sign the new ones and that's it the CLI to automatically handles this so just broadcast that all of this attributes are signed now and that's it. + +Benjamin B: Okay, I understand. But now we're talking about it. You can't sign a specific attributes here, right? but rather + +Andrey Arapov: No, at least like not what I've seen maybe on the protocol layer. It's somehow somewhat possible. But I'm not sure I think the tool just check the real attributes from the chain and just broadcasted. Hey, I signed this sort of but theoretically I think it's possible to broadcast if you modify the code some sort of attributes, I think but the battery harder could clarify it. + +Benjamin B: Yeah, I think maybe this has some potentially using the future, but we weren't planning to use it in that way. We were planning to set up a multiple wallets instead so that if we wanted to sign for multiple things if you want to sign everyone from Germany with one at risk, we can do that for filtering purposes. anyways, thank you a lot and if you want to share A description for that maybe it could be useful for us also. + +Andrey Arapov: Yeah, I'll look at it. Just ping me on the discorder like things. + +Benjamin B: Thank you. And then I want to continue with the last Point here is that with successfully been able to read some of the cloud most API, but now that I read in the Discord that it was updated. I'm not sure if it's outdated or not. But this is a really small thing that this is a fixable + +Benjamin B: Thank you for sharing that link likes. + +Benjamin B: I'll go through this later. So that's everything for the presentation and everything for my questions as well. So now it's up to anyone else if you want to raise your hand and if you want to give feedback or ask questions or anything. Actually, there's one more thing. I heard Tyler in the beginning of King about if we have any deadlines to say because of the way we're doing we're first creating our prototype or our internal solution and then creating the Integrations. We don't have this yet. + +00:30:00 + +Benjamin B: We don't have any rough timelines for anything right now besides and what was mentioned in the proposal. I hope this wasn't your question, please go. + +Tyler Wright: No. Yeah, I just think from a deliverable standpoint. I know that as outlined in The Proposal is a three to six month timeline to do a number of these contributions. I think they intended start date was a end of January. So I think we're close to schedule right now, but I just want to figure out a Cadence and then from a deliverable standpoint when some other members of the community we can go into a testing phase. I know that you have to do some deep diving into provider future discoveries the updated the cloudnos API, so For me, I'm thinking and this is up for discussion right now. If we + +Tyler Wright: Yeah, Anil just dropped in some information about kyc as well. So I know that's something else. I'm thinking maybe for the first three months like a February this next meeting in April. We stay in a monthly Cadence but it seems like after that we should be able to move to maybe a weekly Cadence where we'll have specific things that we will start to be able to see demo and test and continue to optimize that that seemed like a fair enough timeline as we try to hit That six month deadline at the high end. + +Benjamin B: Yeah, I think that makes total sense because most of the things we're doing we can work on all of them separately or most of them but they're not going to get finished at the different time periods. Really they're going to get finished or be able to be tested roughly at the same time. So doing it monthly now works for me and maybe Sebastian has something to add here How you can allocate time for meetings and such? + +rUqeRT: I think it's all good. + +Benjamin B: And that's the same time. as today Wednesdays at 7pm + +rUqeRT: Yeah, sounds good. We're doing monthly and then we're doing by Weekend. Yeah like that. + +Benjamin B: also Is there anything else? + +Tyler Wright: cool No, I would just say maybe as we lean into next meeting you will have more of a plan as a pertains to that first Deliverable look maybe the following month and maybe a short testing plan or something like that. I just wanted to be tracking. A deliverable wise so that folks that may not be able to make the time because again I appreciate I know it's late for a lot of folks in Europe but folks have family. So I want to make sure that if people can't join the meeting they have some specific deliverable that I can hand out alongside the notes and the recording and the transcript for people to track over the next month or two and then we can go from there. With something that's more concrete and visual. + +Benjamin B: Yeah, that makes sense. One last thing for me. is that someone I think it was Anil mentioned how maybe it doesn't make sense to use images to do the benchmarking in that case. I would prefer to do some offline discussion. So please feel free to send at the end if you have some idea how that could look like. But I feel like if the resources are really utilized regardless if you can rent it or you're not going to be able to do it if they're already working the course nourishing. but if you don't reach out and we will probably just continue as normal with our plan to create the docker images to essentially + +Benjamin B: Do some workload to figure out how fast the CPU is maybe it has. test test so that we can actually verify the advertised resources fully and not just how good they are. And then be able to give them scores or a combined score. + +00:35:00 + +Benjamin B: Thank you everyone for joining today's meeting. And thank you everyone… + +Tyler Wright: excellent + +Benjamin B: who could help answer all these questions. That's it for me,… + +Tyler Wright: Okay. + +Benjamin B: but I'll leave it to Tyler if you want to. Have some final words. + +Tyler Wright: No, I again much appreciate Ben walking through I will create a specific repo inside of the community post specific folder for the working group for provider Audits and to be a lot of read me and add the meeting notes for this meeting and then much appreciate everybody that's involved in these conversations today to continue to stay involved be a Discord. I'm going to continue to add a number of folks to the meeting link just so they have it on the calendar. And if anybody wants to be added feel free to let me know but yeah, we'll continue to track the work of the pie three team there again much appreciate the work today and we'll talk more about auditing and that process at the pertains to providers for the incentives at the sick providers meeting because I know some other folks that can make + +Tyler Wright: Here today want to be a part of that conversation. So Alani if you have time next week to join sick providers, that would be greatly appreciated and we can go from there. + +Tyler Wright: All right, I hope everyone has a great rest of their day. Again. Thank you all part three team for leading us through thanks for all the questions and we'll share those resources on discourse that can continue to make sure we're not Doing redundant things as Scott has mentioned that we're just leveraging all the work that the community is doing to continue to push these efforts forward and other efforts so much appreciate everyone's time today, and we'll talk online. + +Andrey Arapov: I think + +Tyler Wright: Thank you. + +Benjamin B: Thank you. + +Scott Carruthers: That's done. Thanks, everyone. + +Luna: Thank you. + +Meeting ended after 00:38:24 👋 +