From 664623f5c7427c881ad2769bf0c36ddb77478e1e Mon Sep 17 00:00:00 2001 From: George Appiah <48031636+iamGeorgePro@users.noreply.github.com> Date: Sun, 14 Jul 2024 08:31:32 +0000 Subject: [PATCH 1/3] Create meeting notes for sig-client #16 Signed-off-by: George Appiah <48031636+iamGeorgePro@users.noreply.github.com> --- sig-clients/meetings/016-2024-06-25.md | 156 +++++++++++++++++++++++++ 1 file changed, 156 insertions(+) create mode 100644 sig-clients/meetings/016-2024-06-25.md diff --git a/sig-clients/meetings/016-2024-06-25.md b/sig-clients/meetings/016-2024-06-25.md new file mode 100644 index 00000000..e2cdf795 --- /dev/null +++ b/sig-clients/meetings/016-2024-06-25.md @@ -0,0 +1,156 @@ +# Akash Network - Clients Special Interest Group (SIG) - Meeting #16 + +## Agenda +- Updates on Human HQ Team's Work +- Updates on Console Activities +- Presentation on Go Client Libraries by Joao Luna +- Demo by Joao Luna on Web3 Platform +- Summary of Near-term Plans for Console Improvements by Anil Murty + + +## Meeting Details +- Date: Wednesday, June 25, 2024 +- Time: 08:30 AM PT (Pacific Time) +- Recording: Coming Soon +- [Transcript](#transcript) + +## Participants +- Anil Murty +- B S +- Deval Patel +- Dominikus Brian +- Jannick H +- Jigar Patel +- Joao Luna +- Julius Haus +- Maxime Beauchamp +- Maxime Cyr +- Scott Carruthers +- Tyler Wright + +## Meetings notes +- **Welcome and Meeting Overview** +- **Tyler Wright:** + - Tyler Wright welcomed everyone and explained the change from a monthly to a bi-monthly meeting. + - Emphasis on updates from various clients and their work. + - Recap of the Human HQ team's demo and their transition to on-chain deployment from GitHub to Akash network. + - Collaboration with the console team, proposal passed on-chain. + - Recognition of the Human HQ team's efforts. + - Tyler mentioned potential changes to the meeting day and time to accommodate global participants. + - Appreciation for attendees joining despite time zone challenges. + - Availability of updates on the product and engineering project board. + - Representatives from the console team (Maxime Beauchamp, Jigar Patel, Deval Patel) available for in-depth updates. + + +- **Presentation on Go Client Libraries by Joao Luna**: + - Joao Luna has been working on the client library for some time. + - The library is part of the core codebase at Akash and is accessible for deeper exploration. + - The goal is to make the library open source, allowing broad accessibility. + - Demonstrated the usage of the Go client library with Akash API, focusing on fee grants. + - Highlighted the benefits of the library in abstracting complexities for developers. + - Provided an example of a deployment using fee grants, shared a URL for demonstration. + - Discussed the potential for integrating memory keychains and other improvements. + +- **Discussion on Leveraging Go Client Work for Akash API by Scott**: + - Scott Carruthers is interested in using the Go client work for further development around Akash API. + - Joao Luna explained how the library simplifies interactions with Akash API and supports various integrations. + - Discussion on managing deployment with different depositors and leveraging fee grants programmatically. + - Scott and Joao discussed potential methods for sharing code and collaborating on documentation. + +- **Demo by Joao Luna on Web3 Platform**: + - Joao Luna presented a live demo of a Web3 platform he has been building for over a year. + - The platform aims to provide a cloud experience using decentralized protocols, offering features similar to traditional cloud platforms like AWS or Google Cloud. + - Users can create their own cloud with high availability setups, select data centers globally, and integrate storage with Akash compute. + - The platform is currently live in a beta state and is open for beta testers to help scale and identify bugs. + +- **Platform Features**: + - Login options include standard login or GitHub integration. + - Users can add credits to their accounts with straightforward billing; no surprise bills as users set a budget limit. + - Marketplace (to be renamed Templating) allows users to create deployments with a user-friendly interface without needing to understand SDL. + - Users can configure ports, container commands, arguments, environment variables, and GPUs. + - Deployment status can be monitored in real-time, with access to logs, billing information, events, and revisions. + +- **Deployment Example**: + - Joao demonstrated creating a deployment using the platform. + - Highlighted the ability to deploy services like Grafana on Akash network. + - Deployment management includes choosing trusted providers and maintaining high availability by redeploying to other providers if one goes down. + +- **Q&A and Discussion**: + - **User's Balance**: Balance is held in USD, not tied to any cryptocurrency. + - **Custodial Wallets**: The platform uses a mix of individual and special wallets to manage deployments. + - **Fee Grants**: With the latest network upgrade, fee grants are utilized to streamline token and wallet management, enhancing operational efficiency. + - **Currency Pools**: There's potential to support stable currencies like USDC for more stability and equilibrium-based systems in the future. + - Joao Luna completed his demo, emphasizing the platform's ability to manage decentralized cloud services efficiently. + - Joao highlighted the upcoming use of fee grants to improve operational efficiency. + +- **Anil Murty's Summary of Near-term Plans for Console Improvements**: + - Anil shared a summary on Twitter and detailed plans on GitHub. + - **Feedback-driven Improvements**: + - **Private Container Registries**: Essential for AI/ML developers who prefer keeping their container images private. This feature has already been shipped. + - **Support for Standard Linux Images**: Mimicking AWS/GCP/Google experiences by allowing users to SSH into VMs with standard Linux images. This will be integrated into Akash to simplify user experience. + - **Automatic Build and Deploy**: Targeting web developers from platforms like Vercel or Netlify who are not infrastructure-focused. This will streamline the process from code to deployment. + - **Custodial and Non-Custodial Wallets**: Supporting both custodial and non-custodial experiences to attract non-crypto users, providing a seamless, non-wallet-based interface. + - **Praetor Integration**: + - **Unified Frontend**: Integrating Praetor’s functionality within the Console to simplify provider management and improve user experience. + - **Storage Solutions**: Exploring integrations with StoreJ, Filecoin, etc., and making these accessible through a user-friendly UI. + +- **Open Discussion**: + - **Questions and Feedback**: + - Questions were raised about the implications and rollout of these new features. + - Anil emphasized the importance of these improvements in removing current barriers to Akash adoption and enhancing the overall user experience. + +- **Further Discussion by Anil Murty**: + - **Periodic Data Backups**: Emphasized the importance of enabling users to easily back up their data periodically, particularly for users who host services that require persistent storage. + - **VM Access Concerns**: Addressed a query from Dominikus Brian about SSH access to VMs. Anil clarified that due to Akash's reliance on Kubernetes, direct VM access isn't feasible. Instead, users can SSH into containers. + - **Micro VMs**: Mentioned the potential future exploration of micro VMs, although it would require significant effort to implement. + +- **Joao Luna's Feedback and Discussion**: + - **User Experience with VMs**: Joao raised concerns about the user perception of running VMs on Akash. He noted that users might face limitations when running Kubernetes clusters within containers, which could lead to confusion. + - **Messaging and Expectations**: Anil agreed that clear messaging is crucial. The primary objective is to provide users with a familiar experience by allowing easy SSH access from their terminals. + +- **Open Discussion and Questions**: + - **Dominikus Brian**: Asked about SSH access and its limitations. + - **Joao Luna**: Suggested exploring alternative solutions for VM-like experiences, considering the constraints of running containers within virtualized environments. + +- **Joao Luna's Final Question**: + - **SSH vs. Shell Connection**: + - Joao Luna: Asked why shell connection isn't an option currently. + - Anil Murty: Explained that users prefer using their own terminals, and there's a bug in the current shell causing session breaks when the provider restarts. It's on the core team's roadmap to fix this issue. + - Maxime Beauchamp: Added that users generally prefer SSH through their own terminal. + +- **Unrelated Query from Jannick H**: + - **Ledger Amino Sign for Akash JS**: + - Jannick: Asked if signing transactions on the Akash console frontend with a ledger is planned. + - Anil Murty: Stated it's not currently planned as it's the first request. Akash JS is open source, and contributions are welcome. + - Tyler Wright: Encouraged discussion on this topic in the Akash JS Channel on Discord for further feedback. + +- **Discussion on Awesome Akash Repo Cleanup**: + - Tyler Wright: Mentioned ongoing cleanup work by PiberDev on the Awesome Akash repo, seeking feedback from the Sig clients. + +- **Closing Remarks**: + - Tyler Wright: Thanked participants, highlighted upcoming meetings (Sig support, Sig providers, and the working group for Akash website and documentation), and urged continued engagement on Discord. + +## Action Items +1. **Joao Luna**: + - Continue refining the platform based on beta testing feedback. + - Explore additional currency support and further integration with Akash's latest features like fee grants. + - Collaborate with Scott Carruthers on leveraging the Go client library for Akash API development and documentation. + - Plan for making the Go client library open source and provide initial access to interested parties. + +2. **Scott Carruthers**: + - Develop engineering documentation for integrators, incorporating examples from Joao's work. + - Coordinate with Joao Luna on code sharing and collaboration. + +3. **Tyler Wright**: + - Continue facilitating discussions and collaborations within the group. + - Consider adjusting the meeting schedule to better fit global time zones and accommodate more participants. + +4. **Console Team**: + - Continue providing real-time updates on the product and engineering project board. + +5. **Anil Murty**: + - Oversee the implementation of new features and ensure they align with user feedback. + - Continue updating the community on progress via Twitter and GitHub. + + +## Transcript From a1be0d221ebc6ec01e522d87f26b371d131087d4 Mon Sep 17 00:00:00 2001 From: Drinkwater Date: Fri, 19 Jul 2024 07:40:36 -0400 Subject: [PATCH 2/3] Update 016-2024-06-25.md added transcript chore: add recording link Signed-off-by: Drinkwater --- sig-clients/meetings/016-2024-06-25.md | 446 ++++++++++++++++++++++++- 1 file changed, 445 insertions(+), 1 deletion(-) diff --git a/sig-clients/meetings/016-2024-06-25.md b/sig-clients/meetings/016-2024-06-25.md index e2cdf795..f4f77657 100644 --- a/sig-clients/meetings/016-2024-06-25.md +++ b/sig-clients/meetings/016-2024-06-25.md @@ -153,4 +153,448 @@ - Continue updating the community on progress via Twitter and GitHub. -## Transcript +# **Transcript** + +_This editable transcript was computer generated and might contain errors. People can also change the text after it was created._ + +Tyler Wright: Alright, welcome everybody to Sid clients. This is now a monthly meeting. It's June 25th. 2024 again this meeting used to be a monthly meeting but due to some teams at some clients that are currently building not having the time to meet and a number of things have been happening with around console the folks at the meeting in April. Talked and we've had some gender more General discussions that we've changed this meeting to being every other month during these meetings. We'll get updates from various clients and all the work that they're doing. + +Tyler Wright: At the last meeting, we got a demo from the human HQ team on the work that they has since been gone on chain for the effortless deployment from GitHub to the Akash network via the cash console. Again, human HQ team led by piersch went through a number of different steps. They gave a couple of presentation's at various sick clients meetings had some pre-draft and just draft discussions got some input from the community met with the console team. + +Tyler Wright: And got some input on how they would make some changes added to awesome Akash as a repo and then again most recently put a proposal on chain for the work that they'll be undertaking over the next six months along with the console team that proposal passed on chain this morning. So again, shout out to all the work on the human HQ team led by Peach and I'm sure there'll be further updates on all the work that they'll be doing with. console team moving forward + +Tyler Wright: Cool, in addition to that again, there's a number of items that I've put on the agenda for folks. So again, thank you to those at had some issue with Discord events much appreciate everyone's time moving forward. I will try to adjust the time and most likely the day of this meeting. Tuesdays is just quite a busy day for members of the product team and core team. So I may change the date of this meeting on a monthly basis and then move the time a little bit earlier because I know some folks that are based in Europe Asia and elsewhere. It's quite late for so I appreciate those folks taking the time to join us today. + +Tyler Wright: All So again I know we have a full agenda one of the things that we'll talk about a little bit later is get an update on all things happening with console for folks that want to see what's going on. I know the console team when we have a number of Representatives here today the Max's here today as well as jigar and Deval if you want to see what's going on again, we'll get some updates. In depth today, you can always track all the various items being worked on the product and engineering project board. Again. The console team does a great job of updating this board in real time. + +Tyler Wright: But before we get to updates on the console again, if anybody has any other agenda items that they want to add please feel free to drop those in chat. One thing I wanted to discuss at the top was go client libraries. This is work that Luna has been doing for for some time and it's been implementing it into some tooling that he'll give a demo on later. But I know that Scott is looking to leverage some of that go client work as he continues to do some work around their cache API. + +Joao Luna: Yeah, thank you. So, yeah, I've been working on the client virus for quite some time. + +Tyler Wright: So I just wanted to see Luna there's an update that you had on go client Library. + +Joao Luna: It's part of our core codebase at Kozak and… + +Tyler Wright: It's accessibility for folks that want to take a deeper dive and… + +Joao Luna: it's something that Indian goal is to be open source. And for everyone I did a demo I think a couple months ago. + +Tyler Wright: then maybe like a quick demo or… + +Joao Luna: I think it's not the one year yet. + +Tyler Wright: quick update all things there. And then I think the next agenda item that we want to talk about is maybe Luna some work that you've been doing on in regards to heart as you maybe want to share. + +00:05:00 + +Joao Luna: It was some time ago since then we've evolved it now. We are using the cash API. So it's something that takes full advantage on the cash API and… + +Tyler Wright: + +Tyler Wright: A Passover. + +Joao Luna: it's compliant with everything there. + +Tyler Wright: Do you Luna to give an update on the go client libraries? And then Scott if there's any initial questions that you had outside of just general access we can take some time here today. + +Joao Luna: and as of the latest Network upgrades with the frequents, our newest features is obviously the support for the fee Grant and… + +Tyler Wright: So Luna, I'll kick it over to you. + +Joao Luna: obviously which allows for A limitless list of Integrations to happen and you can build all sorts of backends and have all these. + +Joao Luna: management of the points much more easily and they are quite thrilled about it. I have a couple of examples on the repository that's using sandbox and I can show you how it works and then we can ask questions. We can enter a little bit of detail and regarding access to the code and all that. Can then discuss further. + +Tyler Wright: Yes. + +Joao Luna: Okay, so this is gold code. Let me know if you're not familiar, but I'll try to get this easier to understand. So as you can see here on the left side, these are all examples of Snippets of code that you can run and test the library is working. As I said, this is combined with a cash API. I know there's plans to build an SDK around the cache API in generate all these apis for different languages. What we are trying here to do here is to have semantic way of using those apis more user friendly for the developer and more focused on the actual. + +Joao Luna: Use cases of the cache Network so users of this API don't really care that we are sending transactions to a blockchain what they want to be able to do is I want to quickly deployment. I want to check my list and have we often interface for them that allows them to do this programmatically without having to worry about Cosmos SDK transactions signing gas fees, whatever. We try to obstruct the most that we can that lead that's possible. Of course. + +Joao Luna: So yeah, we have a bunch of examples. I've shown you before to get deployments or get providers. where you can actually run and get a provider what I want to you to show you today is our latest example, which is basically a full deployment. This goes through the full process of deploying but using the feed grants, so we actually have to accounts on Sandbox one will be the grantor which will basically Grant access and permissions for a deployer to perform actions on it's the health basically a deployed with no way KT will be able to use the APT of the grantor to perform actions and here we do like publishing a certificate creating a deployment and the lease and manifest and also finding the department account. + +Joao Luna: And so before typing is the code what I can show you is basically running the example. This will take a couple of seconds, but we'll have a bunch of logs and at the end you'll see an URL and we'll access and see that it's deployed. And then we can go a little bit through the code just for you to see the structures and what interface looks like to interact with this. + +Joao Luna: Should take a while. + +Joao Luna: Okay, so here we have a transaction. Keep going. + +Joao Luna: I have a bunch of sleeps here. So this is why it takes some time because I'm waiting for the deployments. But now we do get all these and we get a URL I'll share with you and you can check that. It's working. Let me just put we HTTP so you can click it. And you should be able to click and see a deployment. That's basically an nginx. Hello world deployment. As you can see I've also added funds to this deployment and if you want to further check, you can go to the deployer address on Sandbox. I'll share with you and you can check the transactions being done. + +00:10:00 + +Joao Luna: Yeah, so this was all done with basically the funds from a grinder, which was the latest feature that was introduced on the latest Network upgrade. before diving into the code any questions and we want to say + +Tyler Wright: Go ahead. Don't me. I see your hand. + +Dominikus Brian: Okay, I have one. So with the latest features in console, we can choose which Grant right in this case of yours. We can also choose. By giving these different string. + +Joao Luna: of course. exactly, so + +Dominikus Brian: And can you have a list of them and then just pick? + +Joao Luna: basically you say for instance. Let's say here when you are finding a deployment. One of the things that you say you have to pass the positor so you can't really have a list what you can do is have your own list and then accordingly just choose a depositor based on that. So you can do that programmatically. It's not supported by the API with something that perhaps doesn't make sense for everyone. So we try to offer an equilibrium between what's simple and makes sense. And what's useful. + +Dominikus Brian: nice probably + +Joao Luna: But you can specify a depositor, of course on the funding of the deployment and on the deployment creation as well. You basically do let me see here as well. So the create deployment options also added a depositor address. + +Joao Luna: Yeah. Yeah, So this depositor don't need to be the same as long as you have you have for the address that's deploying. It's all fine. And that can be seen being done here somewhere Health secrets basically are saying that we have this granted address this the point address and we are setting a spend limit of 1880 and with an expiration of one year. + +Joao Luna: from today + +Joao Luna: That's our goal. Yeah. Thank you. + +Tyler Wright: Thank again access I know this is something that is currently not open source, but there is an intent to make it open source of more people that can interact with it. Is there a plan for that or can we take initial steps to give some people access a Scott and then I just want to follow up Scott that you have any questions that you had at this time? + +Scott Carruthers: Yeah, no, I don't have any questions + +Joao Luna: I saw your message about the access. My only concern is that because it's a private organization. I don't know if I have to pay for a seat to give access to someone to do that private, but that's something we can take offline and discuss and in the limits of things we can just go through the code on a call run share the ID and have a remote developing session where Scott can try. Yeah, I'm open for finding kind of solution. + +Scott Carruthers: yeah, so as the more I think about it Luna I could just reach out to you directly and it could probably just be as simple as if you just wanted to share something and adjust or Even just in Discord provided to me directly. So let me Take a step back just to give you some insight. The reason for my current interest is. I'm going to start developing. More engineering documentation for integrators. So we want to make it very frictionalist to Interact with a cash card base and Akash API. So this won't be a matter of creating a client Library. It would more be a matter of bringing them through examples of how they would interact with the API and I just want to as I go through constructing some of that if we were using some more strategies than some of that documentation would + +00:15:00 + +Scott Carruthers: Yes, some of that documentation would also speak into what you so it just said as an example of creating a deployment. I'm sure we would use very similar techniques but it would be good. If I just had inspection of the code that you're using and what I would suggest then documentation and try to get some cohesity there. So as I'm going through that and I'm going through a creative deployment example, I can just reach out to you and maybe you could just share with me directly. They create deployment example, and that's probably fine for now. + +Joao Luna: Yeah, trying to make sense. Some of the problems that we solved are also not really related to Akash, but more on the cosmos SDK and how we can import Keys all these keychains because we are not using operating system keychains. We are using in memory and those have some quirks when you are Building Services. You don't really want to use the OS ones you perhaps want to use the memory keychain. There's some quick works that we need we had to do to get it working some importing methods that we needed and we solve those problems or perhaps we can cooperate that and then if it's open source, it's just like the standard way of doing things we can agree on. + +Joao Luna: But yeah, so this is that the more complex example, then I can also just show you running something very simple you can do is just for instance querying providers as you can see. This is the cache API. So it's fully compatible with the cash API. We didn't want to introduce more complexity because if we think that we want to open source this and move this to the cache Network organization. We don't want to offer a different API with a different set of data interfaces than the developers of the Akash Network people that use the API will be used to so we decided to keep the same abstractions, but just offer more semantic way of interacting with the network some way that's cleaner makes more sense and has no blockchain quirks here and there so we try to abstract not just possible. So for instance getting a specific provider is + +Joao Luna: creating a request and then just get provider with the request. This is very very simple example and books very clean. And gets you Another thing that I want to point out is that this is a go Library that they're not really code generated from a set of protobufs. We have some control over the interfaces that we provide and we can do some useful or some more like standard or cleaner way of doing things. So for instance when you are creating a deployment. + +Joao Luna: One of the things that we do is accept IO reader when we are building the deposit the message. So what this allows is, we are offering a standard Library interface our Channel way of using this library and this sdl reader can come from any kind of IO reader. This can be a file Network socket a stream whatever you can basically deploy as long as it's respect this interface, which is standard gold length stuff. So it opens a lot of possibilities. So inside of reading a nasty out from string and passing it through + +Joao Luna: Is just whatever you have a reader is supported so you could be receiving streams of sdls and creating the partners. Theoretically if that's your use case, but this is the design that we had in mind is that's make this the go way and… + +Dominikus Brian: I have one last one. Perhaps pretty simple,… + +Joao Luna: then interact with the cash networking that go away. Kind of thing. + +Dominikus Brian: but when we have several of this clients for instance trying to deploy using the same address and… + +Joao Luna: And yeah, that's pretty much it. + +Dominikus Brian: everything parallel operations will there be conflict? + +Joao Luna: There's one limitation but that's solved in the new Akash API and solved in this scope client. So basically imagine you sent two different. Operations to the network not queries, but transactions programmatically in parallel. They would to be two different transactions with basically the same sequence number and… + +00:20:00 + +Dominikus Brian: I see but in the case of deployments for instance. I want to deploy us the hello world example you have… + +Joao Luna: that's not allowed you can only send one per + +Dominikus Brian: but to three people instead of one,… + +Joao Luna: One per block. + +Dominikus Brian: so I'll want to deploy three times together. + +Joao Luna: What you can do is basically send an aggregate operations and… + +Dominikus Brian: You could and… + +Joao Luna: then send multiple messages on that same transaction,… + +Dominikus Brian: you do it better inside, one program or… + +Joao Luna: and that's currently support. + +Dominikus Brian: you run several of them together. Because running several might be problematic right because they're using the same address. They are trying to + +Joao Luna: You could. + +Dominikus Brian: + +Joao Luna: As long as you keep when here I'm not sure we are entering a realm where you have these parallel or concurrent stuff. It's not clear for me. + +Dominikus Brian: The first option I guess safer. + +Joao Luna: Perhaps Scott has more experience or… + +Dominikus Brian: Yeah. if just looping together and… + +Joao Luna: has dealt with this because I did multiple transactions,… + +Dominikus Brian: then deploying one by one because from your demo just a few seconds,… + +Joao Luna: but I never did multiple deployments. And I think makes sense. + +Dominikus Brian: so you can actually + +Joao Luna: It would work as long as you're sending the messages in the sanction transaction. There's no problem. You don't have to wait for another block. To be able to do that. So I can't see that happen again. + +Joao Luna: Yeah, but It What you ask makes sense and I got into the issue on the terraform provider client as well because I was deploying my infrastructure with the telephone provider and at some point I was doing parallel just to optimize and we were entering conflicts because we are doing many transactions on the same block which was allowed. + +Joao Luna: But that's something that initially you can encounter… + +Tyler Wright: I do want to leave some time for again a short demo from Luna on… + +Joao Luna: but it solves here because the transaction support multiple messages and it's pretty easy to do that now. + +Tyler Wright: what he's been building and then dive into all the work that has been going on around the console 2.0. + +Joao Luna: And that's thanks to the new Akash API code base. + +Tyler Wright: Does anyone have any additional questions here I saw that you were unmuted for quick second. + +Joao Luna: nice + +Tyler Wright: Didn't know if you wanted to chime in or if we just want to move on. + +Tyler Wright: Perfect. Thank you. + +Tyler Wright: Yeah, yeah. + +Joao Luna: Okay, I guess then it's the other demo. I'll try to be quick about this one and hopefully it works. It's a live demo. So I'll share my screen. + +Joao Luna: Yeah. + +Joao Luna: Yeah. + +Joao Luna: So very quick very briefly. This was a platform that I was building for one year I think on my free time but things will just accelerate and we were having fun with it. So we kept building the idea overall is to have a platform that increase a bunch of web free on the deep in protocols and offer the regional experience traditional Cloud platform experience, but where you and your Enterprise can build your own cloud kind of thing. + +Joao Luna: and that their own called platform tailored to your needs where you can have your high availability setups. You can have your nodes or your data centers that you choose across the globe. You can integrate storage with cash computes in various ways. That's kind of the vision is you build your own cloud and you have your traditional Cloud platform that you're used to AWS Google but using all these decentralized protocols and basically have it more like educative and so Enterprise grade kind of features such as high availability load balancing and all that. so yeah, currently it's + +00:25:00 + +Joao Luna: I'm running this on a local environment it is life, but it's open for better testers because we don't want to have many people accessing. just yet just to scale properly and not planning to many issues right away and + +Joao Luna: and ruin the user experience. So it kind of like uncovering a bugs here and there are experiences that are not that great. But this is like a brief landing page, but what's really interesting here that can basically log in or you could use GitHub as well. It's already supported. It's quite the easiest way to + +Joao Luna: to go about it, but I already have my account. I also already have balanced. So I'm not going through the process but we want to have the most straightforward way of adding credits to your account. You basically pay as you go you have your credits. You don't get surprise bills like you have to pay $1,000 because you left the deployment here. It's like you reached your limit that you set from the budget. Then the deployment gets close you get of course warnings, but it's close. You don't pay overpay or whatever but the billing page we are still building. This is very better but I can show you here creating a department basically and this kind of like Marketplace Michael places that name we are trying to change to templating and one of the things that we have currently is basically and hello world engine X is the same example that I did before so why it does he's kind of prayer feels these four basically have a + +Joao Luna: Choices here you can go for your ports. It can have container command arguments. Let's start our stuff that you are used to add variables the gpus with and as you can see it's a very traditional experience. So forget the sdl that's not something everyone we want users to worry about is you are kind of like in Google Cloud platform or Hefner and they're creating your own server. So you feel this and you basically just create. And you have your deployment then we just need to wait for it to the point. So it shouldn't take long. I'll just keep refreshing and see and we can see. + +Joao Luna: In the meantime, what I can do is show you this page, which is basically the page that will allow you to overview what your D cloud looks like so basically the group of providers that you trust in shows for your own cloud and this is kind of like the page. So I chosen for the form for this demo and currently the Europe plots The Foundry digital this in North the overclock and the three Akash Cloud providers. Then I can have more details on the providers. it shows you a page with attributes. We have as Telemetry not in this environment. You'll have all sorts of data. This is these pages of course being worked in. It's kind of raw still. + +Joao Luna: But yeah, We have the points status. Let's see. Okay, I'll click logs. Hopefully this works. Sometimes it doesn't so you have access to your logs. You basically have a bill this is your active bill for that deployment. What it says basically is that how much you'll pay it pays every hour and this is wrong you pay every hour. And yeah, that's pretty much What can happen is if your deployment on your provider goes down. + +00:30:00 + +Joao Luna: you can choose for you to be automatically deployed to another one and you'll just have a new bill for that provider. So you can keep up to date with the pills. that are active at the moment. The other one would be inactive and the new one will be you then have events what happens so everything went smoothly revisions is the updates on deployments. So if you update your image, you'll be able to see when did you upgrade that image? And what's the current one? So this is what we call revisions. It's the same kind of language that you called platform uses Google Cloud. So it's deployed the four parts none. So we have where else and then I can click it's being hosted on your plots. And we're getting service department available. Not sure matter if it's + +Joao Luna: airport error. It's not there for 04. So, let's see if I delete it. I should see it 404 + +Joao Luna: eventually + +Joao Luna: Not sure if something's wrong with the robot provider. + +Joao Luna: it's offline now. they all need to check what I can try to do and I think that's pretty much You can ask questions different also have documentation. In the meantime. When you gather your questions. I'll try to show you an example deployment, but feel free to ask any questions. + +Joao Luna: JD is to have this as an enterprise-oriented solution for the point on a cache and other. Solutions it's currently we only support the cache and for compute that's our main focus really. We haven't considered any other options, but for storage we are looking for other networks and then offer a layer of integration between those networks and offer some Enterprise solutions that you can't get from being like let's say a layer one platform on a car. + +Tyler Wright: There's a question in the chat. + +Joao Luna: So for instance that some limitations that Make sense that a cache network has that can only be overcome… + +Tyler Wright: It's a user's balance in usdc or Akash on chain, or is it not and + +Joao Luna: if you have a platform built on top of it. that has the higher level set of features. So basically you kind of like sacrifice control. Over offering this features. What we want to do is give people the control to Be able to make the informed decisions on what they want to sacrifice in terms of. Fine-grained control and get good features like high availability. But yeah, that's kind of like the + +Joao Luna: the bridge that we are trying to build kind of thing. Let me just check if the funny is working. Okay, so the professional deployment is working perhaps it's the other image. + +Tyler Wright: go ahead and you? + +Joao Luna: So yeah. As you can see I can log in and I have graphana deployed on a cautionary. So wallets, just The traditional product experience you would expect from a cloud platform. + +Joao Luna: It's nothing it's our own. it's your dollars, basically. It's like having dollars in your bank account, I think. It's not tied to any cryptocurrency. We don't buy cryptocurrencies with your fans. So for instance. it's possible that if you purchase credits on our platform and for some reason it wanted to refund it's your dollars so it would be You don't purchase crypto. It's just you share this link with a friend. That's never heard. + +Joao Luna: What sorry what's used for the appointments underneath in your custodial wallet? It's the cash go client library that we are building. So these are being built and in. + +00:35:00 + +Joao Luna: Does that answer your question? + +Joao Luna: First thing is that we don't use just one wallet. So we have our own separation of + +Joao Luna: Deployments let's say the point agents or wallets what the currency that we are using we currently have a system where you have a bunch of pools of tokens. We currently have one active pool of akt but ye is to support whichever currency are supporting for us. I think it would even be better to have perhaps who DC. I don't know just have a more stable pool. Perhaps can do some equilibrium based something real developing. + +Joao Luna: a Deval + +Joao Luna: Okay, and you can leave your questions in the chat as well and I can still answer. + +Joao Luna: So currently what we have implemented is a mix so every user has its own wallet, yes, but then we have a couple of special walls that serve some purposes And one of the things that we are building now is because of the latest Network upgrade. We really want to take advantage of the feed grants. And that will help a lot on the operations that we do behind the scenes. We have a bunch of services that have to manage all these tokens and wallets and with this picture that's basically gone, which is great. So that's why I'm really happy for that feature. + +Anil Murty: Thanks, Hey everyone. So actually did a quick summary of this on Twitter. I don't also put a more detailed version of that in GitHub as well. But I kind of wanted to highlight what the near-term plans are as far as improving overall experiences and console is so the link that I put on Twitter. I just kind of summarize that in this document as well, but for the new term objectives everything that we're going to be doing is Driven by the feedback that we've gotten from many of the pocs that have done with customers. We have a whole bunch of videos. He's going on you probably heard about the integration of flock you've heard about the integration with neural AI with brev with the whole bunch of these customers and partners and what we've learned from that is four or five things. And so that's kind of summarized here. + +Anil Murty: The first thing that we learned was that a lot of AI and ml developers today. We just kind of the group of people that are driving a lot of demand for gpus right now. They like to be able to keep their container images private and so supporting private container Registries in the core of the product as well as within console for deployments was kind of number one and that's actually shipping now. So it's shipped in it's available. The second one we realized was many of the customers and partners that we speak with particularly on the web to side. They come from products like AWS or gcp or Azure Deputy public clouds, even tier one or two public clouds and the most common workflow there obviously is like if you're going to gcp today and you spinable VM what you get typically is the option to pick a given Osage. So typical Linux image, you're going to LTS 24 version 24 or + +00:40:00 + +Anil Murty: Your red hat Enterprise Linux or one of those and typically users are able to pick one of those options configure the hardware that they want in terms of CPUs memory storage all of that. And then what they end up getting is a way to SSH into that virtual machine and be able to download any of the applications are any code that they would like whether it's models or data or pytorch or anything really any Frameworks that they want to develop with. So people expect that sort of an experience when they come into our cash as well. But what they get is basically experience right now where we expect people to be able to containerize the image and put that and be able to work with that so offering something that is closer to, playing Linux images kind of number two, and so one of the things you'll see on the project board that I was talking about is we're going to be building your workflow where we offer, really build images for the most commonly. + +Anil Murty: Bistros and then give people the option of either dropping in their public keys or generating a public and privately combo and then using that public key to essentially give them access to SSH from their own terminals into this container image. That's a plane relax image and then the user canvas the access HT into that container image. They can download any applications. they can download by torch whatever they like. So that's kind of number two. + +Anil Murty: Number three is automatic build and deploy. so this was finally targeted towards people that are infrastructure Focus people coming from AWS and gcp and Azure. This next thing is targeted towards people who are coming from web to products but are not infrastructure focused. So these would be typically people that are fewer web developers who have never worked with direct infrastructure. I've never spun up, VM in AWS or gcp. They typically work with platform as a service products so past products like for sale or netlify that essentially abstract away. + +Anil Murty: The underlying infrastructure and let people just be able to deal with code. So if you're a web developer and you go into something like a versal experience What You Get There is the ability to just have your react application sitting in GitHub or gitlab or big Market or one of the Version Control Systems be able to just connect that to us and then have whistle pull all the code build it for you and then give you a deployed version of that application. So what you don't have to worry about is Bill process and the infrastructure has been a process. So this is kind of number three and this would be very key to attracting these specifically web developers who don't want to deal with such as so this work is all being done by the human HQ team and there's a proposal on chain that just passed this morning that they are going to work on implementing all of this gonna console as well. + +Anil Murty: The fourth one is the most obvious thing that we have talked about for a really long time, which is supporting both ial as well as a non-custodial experience for people that want to use So today everything is obviously + +00:45:00 + +Anil Murty: Non-custodial meaning everybody brings their own wallet and connects to console using that what we are going to be working on is a custodial experience where we would maintain multiple custodial wallets and have people use those for deployments or rather deploy, their workloads discuss your words, but what the ultimate user sees is a non-volent based experience meaning a non crypto experience. So that's number four and this is going to be key to driving adoption Beyond crypto audiences and… + +Tyler Wright: Perfect. Thank you Anil. Does anyone have any additional questions for Anil or… + +Anil Murty: then the fifth one which is sort of a little bit unrelated to deployment developer experience… + +Tyler Wright: some of the other folks here that are currently working on all this stuff related to console 2.0? + +Anil Murty: but is key to helping us brain into more cohesive experience a brand as well as utilizing a lot of the common functionality right now exists within two different apps. So obviously pretor is the app that many providers use today for spinning up a providers on Akash. + +Tyler Wright: Go ahead go don't mean you go first, and then we'll have Luna follow you. + +Anil Murty: Yeah, that's a really good question mi. So I said, this is very near term objectives, the prioritizing things that are causing people to not use our cash today, but once we get through these there's a bunch of other things that we want to look at in the future. + +Anil Murty: A few months ago, obviously the pretor team joined the overclock lab team. + +Anil Murty: So storage is definitely one serials… + +Anil Murty: And so what we're going to be working on is a new frontend for pretor that is going to decide inside console and… + +Anil Murty: where today it requires a significant amount of detailed technical understanding to be able to set up Akash to be using store J. + +Anil Murty: it's going to be available through some kind of a UI switch… + +Anil Murty: For example, or file coin or… + +Anil Murty: where given provided can come in can switch and… + +Anil Murty: SD. So the next step of this would be exactly that… + +Anil Murty: then, manage their provider and… + +Anil Murty: which is adding to the developer experience so that once somebody has set up or… + +Anil Murty: then switch back to the deployment console and then deploy so that same thing so it's a very similar user experience to something that you would see in other two-sided marketplaces. + +Anil Murty: has a workload deployed. We want to make it easy for them to be able to periodically back up their data… + +Anil Murty: So, if you take a B&B is example… + +Anil Murty: because … + +Anil Murty: if you're hosting a B&B and… + +Anil Murty: even if you have a student storage on a cash today,… + +Anil Murty: you want to … + +Anil Murty: which is supported in console today,… + +Anil Murty: go and Lease in AP yourself, you can do that from the same application. + +Anil Murty: you can figure persistent storage. + +Anil Murty: So very similar experience to that is going to be available within console and… + +Anil Murty: The challenge of course is that if your lease was too close for some reason then you lose all your data that store on the closest storage even the technically + +Anil Murty: any of the common features that exist between pretor and console today whether it's a provider dashboard whether it is the uptime checks,… + +Dominikus Brian: Okay, thank you. That's great a short one relating with the point number two on the vm1. + +Anil Murty: all of those will get merged together. + +Dominikus Brian: so Can we SSS Instead of to the docker… + +Anil Murty: No. that's right. + +Dominikus Brian: because that's more making sense. + +Anil Murty: Yeah, unfortunately So obviously because Akash is entirely dependent on kubernetes as the underlying orchestrator for workloads on a given provider and deciding which node runs on we can't provide people to pay medical VM access. There are certain Solutions like micro VMS that we have thought about in the past, but that's a significantly more lift for us to be able to operate on It's definitely something that we look at probably in the future. so those are kind of the five main things in terms of new term objectives. So these I'm talking about are probably in the next three four months the overall plan is to try to get in a experience But if you have any thoughts or ideas around it definitely open to discussing those. You… + +Anil Murty: + +Anil Murty: + +Dominikus Brian: Okay, great. Thank you. + +Anil Murty: experience that looks and thank you. + +Joao Luna: Yeah, my question was around that so I'm not sure… + +Anil Murty: what like this which is regardless of… + +Joao Luna: if I have much to have. + +Anil Murty: where the user comes from whether they're,… + +Joao Luna: I know I thought without the solutions such as cubert that could be used… + +Anil Murty: crypto user or non crypto user whether they're coming from a infrastructure as a service like AWS or… + +Joao Luna: but my concern was over the idea that we are kind of offering a VM experience. + +Anil Murty: gcp or… + +Joao Luna: That's not really a VMware still connecting to a container. + +Anil Murty: the coming from a platform as a service like versatile Phi all of them need to be able to work with Akash to be able to ultimately get a application deployed in SSH to be able to get into that specific container. + +Joao Luna: There are some limitations. And I was wondering if I wouldn't be a good strategy It start perhaps exploring that way of that new service to provide a jazz VMS. + +Anil Murty: And be able to do that both to a crypto wallet as well as an encrypt over it. + +Joao Luna: I think even for machine learning workloads. + +Anil Murty: So that's kind of the main near term objectives. I don't talk to a bunch over there. + +Joao Luna: It's better to have a VM running. + +Anil Murty: Happy to take any questions. + +Joao Luna: I'm not sure how that works with keywords. It's kind of still running it on a container, but I think it's kind of like a very modified container. That's basically a VM, but I'm not + +Joao Luna: 100% comfortable with saying it's a real VM. But I know as long as you're running on bare metal kubernetes cluster, it's basically the same as running APM things get complicated for running on a visualized environment like running your provider on a virtualized machine, then it would be running a VM over a VM and that's Not like the ideal. + +Joao Luna: But yeah, I mean, I think it's great that we offer that experience. it's Game Changer, but we are just kind of abstracting the problem right? You're still connecting to a container. + +00:50:00 + +Anil Murty: Yeah, you're absolutely Yeah. So this is definitely more of a short term or rather the first step solution for making people feel like they're getting a familiar experience. We have definitely had multiple instances of users who come in, they understand it and then the I can work with this part. Is there a way for me to just SSH into the Container rather than having to do it release shell and this at least Yeah,… + +Joao Luna: Yeah, I understand. The thing that is very interesting discussion. + +Anil Murty: pretty absolutely right. + +Joao Luna: Sorry because for instance what we can end up. discovering is someone saying Okay, I'm running this VM on a card, but I'm not actually able to run let's say a kubernetes cluster that because you can really have Corona kubernetes on a container so we can start running into this power problems if people are confident that they're running on a VM and then To start to work to want to deploy things on those VMS. So in Docker and stuff like that, but it's a very interesting problem. + +Anil Murty: That's Yeah, exactly and I think part of it is going to be our messaging also the primary, pain point to be solving here is not necessarily provide. a little bit of it is making it feel like they have a experience. But the primary objective here is to let people be able to easily SSH from their own terminals as opposed to having to use Alicia. So + +Joao Luna: That leads to me to one last question. It's kind of technical but perhaps what two million answer is enough. Why isn't the shell connection an auction? currently + +Anil Murty: two reasons one is people just like using the terminal and second there is actually a bug in the currently sell which basically causes delicious to session to be broken when the provider part restarts. + +Joao Luna: Okay. + +Anil Murty: There's actually an open issue for that as well. But yeah, it's on the core teams roadmap to fix but outside of that many people just don't want to have to do it. If you are time. There's different about terminal as well. Thank good. + +Maxime Beauchamp: Yeah, and also users prefer to connect to SSH through their own terminal. Or most of the users. Just wanted to add that on top of it. + +Anil Murty: Yep. That was the primary here as a primary. + +Joao Luna: That makes sense. + +Tyler Wright: For all right. I know we have about five minutes left jannick put a question unrelated, but with asking I know that currently it's not possible to sign transactions on the cash console front end with a ledger his Amino sign plan for a cost JS. I don't know if that's something we can answer now, but I just want to go ahead and + +Anil Murty: Yes, it's not currently planned simply because this is the first time anyone has asked for it. At least that I know of but akashias is an open source code base. So if somebody wants to take that on and if you feel strongly enough about implementing it then definitely, happy to support you online. + +Tyler Wright: Yeah, it please feel free to drop some discussion points around this in the cost JS Channel and Discord and then some folks that may not be here today will look to get back to you as well as some folks that are here today, but much appreciate you bringing that up. Any other discussion topics that anybody wants to talk about right now? related to Sid clients + +Tyler Wright: Cool, before we go again. We don't have to talk about it and I dropped it in Discord, but this is something just pybrid Dev was previously working on some cleaning up of awesome Akash repo. I just wanted to have people that at least take a couple seconds to look at it and get some feedback. I know that awesome akashes underlying for some of the work that happens inside the templates for console. So again, this is something that hybrid a did it on his own and I've just been getting some feedback from democus, but I want to get some feedback from people in the Sid clients, more publicly. So again, if you have any comments, feel free to drop them in Discord, and then we can just follow up there. + +Tyler Wright: Cool, if there's nothing else then again much appreciate everyone's time today Appreciate Luna and Anil both going over what's going on a console front as well as what's going on with go client libraries and some other work that Luna has been working on for a long time. + +00:55:00 + +Tyler Wright: One thing I'll also add again. I thank you to all those that bear with the issues of Discord. I please always use caution Network website is sorts of Truth. We have six support coming up tomorrow during that meeting where we might be talking about restructuring of six support and maybe a new Sig that may come from that. So again, feel free to join that. If you have anything that you want to talk about relate to the core product and some of the open issues whether they be related to again the core code base or console. We also have sick providers monthly meeting and then the working group for their cash website and documentation on Thursday and the steering committee as well. So again, we won't go into great detail about number of things that we talked about in detail today during this call, but much appreciate everyone's time today, and I hope everyone has a great rest of your day. Feel free to again stay engaged between meetings in Discord. + +Tyler Wright: Big clients to score Channel or jannick you can feel free to ask any questions related to akashias there and we'll continue to talk. + +Tyler Wright: excellent All right, Max. I'll follow up with you about a piper debt. But again, please feel free to drop any comments related to awesome Akash in there. And again, I hope everyone has a great rest of their day. Thank you all for your participation today. Talk to you all soon. Bye. + +Meeting ended after 00:58:42 👋 From 68b989047136dee4d5a69a61197ce9a3b7e5df10 Mon Sep 17 00:00:00 2001 From: Drinkwater Date: Thu, 25 Jul 2024 12:54:30 -0400 Subject: [PATCH 3/3] Update 016-2024-06-25.md added recording link Signed-off-by: Drinkwater --- sig-clients/meetings/016-2024-06-25.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/sig-clients/meetings/016-2024-06-25.md b/sig-clients/meetings/016-2024-06-25.md index f4f77657..6ecd5373 100644 --- a/sig-clients/meetings/016-2024-06-25.md +++ b/sig-clients/meetings/016-2024-06-25.md @@ -11,7 +11,7 @@ ## Meeting Details - Date: Wednesday, June 25, 2024 - Time: 08:30 AM PT (Pacific Time) -- Recording: Coming Soon +- [Recording](https://g4sync4igp5y56x5cxfhh45vo4elgdzmcptwja4z33gcjhzkkoqq.arweave.net/NyWGi4gz-476_RXKc_O1dwizDywT52SDmd7MJJ8qU6E) - [Transcript](#transcript) ## Participants