Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

What is preventing me from actually using Akash? Server resets. #276

Open
bitcoinmeetups opened this issue Jan 18, 2025 · 11 comments
Open

Comments

@bitcoinmeetups
Copy link

What is preventing me from actually using Akash? Server resets.

So what I have tried to do many times is launching those Ubuntu SSH servers. I install a few packages, fine. But then at some point invariably the server resets. Don't tell me to use the persistent storage because I still have to install all the packages all over again.

Problem is you never get any warning that the server is about to reset. Or any idea why it happened.

Is it bandwidth?
Is it packages not being allowed?
Is it something else?

This has to have a solution for Akash to actually be usable.

@jimaek
Copy link

jimaek commented Jan 31, 2025

For me the whole thing is completely broken. Nothing works. Even though on paper this is the perfect fit for my use-case, hosting a docker container in every available location to power https://globalping.io/

Maybe I just don't get crypto projects

@bitcoinmeetups
Copy link
Author

bitcoinmeetups commented Feb 2, 2025

No, you're not alone. I think akash is more of a proof of concept than a real tool. I have tried to use it but it just doesn't work.

Thing is this is what happens - some guy like the IPFS founder for example talks about his vision and big ideas. Starts building web project. Then realizes hey the government won't like this. (if IPFS actually worked and/or actually were in use it could also be used for torrenting/piracy). But then his project is big and he's making money. So he compromises everything - still no proper search engine for IPFS after ten years for example. And then keeps going whistling as if nothing happened. Now he is the opposite of useful. People who want to do something like IPFS will see oh there's already a project doing that. Let's work on that. But these types of projects are stymied already. The efforts won't go anywhere.

I think it's the same with Akash. It could be useful. It could also be something the government doesn't want - decentralized servers? Woah. So they self-stymie the project. Now Akash looks like it's a decentralized servers projects and attracts attention as such. But it's not. It's not what it appears to be. There are hundreds of projects online like that. Social media for everyone! Facebook. But Facebook is not what it appears to be.

There should be a word for these types of projects that appear to be one thing but in reality they're not. Because if they did operate the project properly the government would lock them up or there would be some other frightening controversy or they wouldn't make as much profit. But they don't want to let go of the project so they keep operating, keep on whistling as if nothing special is going on. Fake attention seekers you could say. I'll think of a better term.

@bitcoinmeetups
Copy link
Author

bitcoinmeetups commented Feb 2, 2025

I'd say Akash usually don't reply much to issues. I would be happy if they proved me wrong though. I'm not saying I'm 100% certain Akash is not going anywhere. But at the moment it looks a bit like a nothing burger or worse - one of those projects that falsely appears to be what it's not.

@cloud-j-luna
Copy link
Member

Hey!! The resets are a real pain, I agree! But what you are trying to do with VMs is not in the nature of the network right now. Yes, the network has a problem with one of its Operators deleting leases or providers requiring frequent maintenances, but the individual provider is as reliable as its Kubernetes cluster which is one of the biggest platforms for compute in the industry.
If we are talking about Serverless Containers, Akash is a great product, arguably superior to its centralized alternatives. Its value, IMO, lies in the potential of new features, in part due to the cloud-native community and all the projects around it. For your specific case, VMs require a little bit more than what is offered, as they would need a new type of deployment and integration with a product such as KubeVirt which implements this rather nicely. The tech is there, it's missing design and implementation.

We are rather responsive and encourage a lot of these discussions. We developers, better than anyone, know the limitations of the network and are working on how to solve them. UX improvements are adopted but don't necessarily implemented straight away due to prioritization or the community building their own integrations with Akash. As an example, fiat payments were first implemented in community projects and later on the official console. My point is that as an open-source project, features and improvements are tackled by different teams that believe one way or the other in the potential of the project. This wouldn’t be possible with non-responsive teams.

Akash’s infrastructure is decentralized. What exactly are you pointing out as centralized in the network? I can think about a few “centralized” components in the ecosystem but would like to hear more from your side!

PS: for this discussion to gain more traction I would start it in the GitHub Discussions forum.

@bitcoinmeetups
Copy link
Author

Hi. I appreciate your reply cloud-j-luna.

By the way, is there anything fun I can do with Akash now then? My way of operating is normally I just start working on servers, then I install packages while I research what I'm doing and how to do it. So it might differ from those that have specific server setups already in mind. I'm more of a bazaar of packages that grow day by day as I polish my devices and make them run more efficiently and frontier expandingly. (New term there).

If you know of anything I can do at all to let me just install as I go, even if there are limitations, kindly let me know. I have tried to use Akash for a long time now but those server resets are just killing all enthusiasm I had for this project.

Feature request, high priority:

You MUST HAVE some kind of warning system before the servers resets AND/OR some way to not make them reset at all. How do you think the users feel when we put tens of hours or days into initial configs and then everything is just wiped out. Rhetorical question so I'll skip the question mark.

The platform itself MUST HAVE built in protection from resets/wipe-outs. Possibly some kind of redundancy. Otherwise it's just completely unreliable. I mean users will stay away from the product. The worst thing that can happen as a server provider is allowing the customer's servers to reset. This MUST NOT HAPPEN. I understand the value of your token is growing. But THIS SHOULD NOT BE THE MAIN POINT. The main point should be to provide a reliable server service. As I said there is no warning, no explanation, no replies to issues, nothing.

Also perhaps reply to issues here within days instead of within weeks or not at all. If you have a link for the github discussions - I assume there might be several - please post it here.

You can also hire me for a while as a kind of tester or something because I actually like Akash, theoretically, if it would actually work (for me). And I'm not expensive. I understand the feelings of the users. There are several apps in the Android play store that have functions which I have suggested, as a kind of resume.

@bitcoinmeetups
Copy link
Author

bitcoinmeetups commented Feb 3, 2025

Actually and I hope this is not too late to implement, Akash should perhaps strive for both built-in redundancy and zero-knowledge/encrypted servers. Have you considered teaming up with Yggdrasil somehow?

Without zero-knowledge, server providers might get pressured to shut down their servers sooner or later. So consequently Akash might not in reality be able to scale. And if it's only about the value of the token without realistic plans to be able to scale then it's not as fun as if I were actually able to use it for something as a customer.

@bitcoinmeetups
Copy link
Author

Update:

I checked out the current Akash ecosystem at:

https://akash.network/ecosystem/deployed-on-akash/showcase/

Looks better than last time I checked. Hope it's going in the right direction. And if I could have some way to work peacefully in an ssh environment in a kind of sprawly way like I usually do then that would mean something.

@cloud-j-luna
Copy link
Member

By the way, is there anything fun I can do with Akash now then?
It depends on your definition of fun.

  • You can run open source LLMs.
  • Stable Diffusion and generate images.
  • Your personal website
  • Testing databases
  • Game servers (beware of the non-redundancy issue)
  • Your own VPN
  • ...

My way of operating is normally I just start working on servers, then I install packages while I research what I'm doing and how to do it. So it might differ from those that have specific server setups already in mind. I'm more of a bazaar of packages that grow day by day as I polish my devices and make them run more efficiently and frontier expandingly. (New term there).

I understand your use case. Take no offence in what I'm saying but perhaps you could see this as a way to improve your current design. There is a flow in your design because a server reset (which can happen on any Cloud Provider) is enough to make it unreliable. Perhaps those packages you mention should be described in some sort of bundle file that gets loaded in your docker image and loaded at startup? This way you can develop your own redundancy and rollout your changes (for example, new packages) across providers automatically. Containers are designed to be ephemeral. This is just food for thought. 😄

I have tried to use Akash for a long time now but those server resets are just killing all enthusiasm I had for this project.

I'm sorry you feel this way, rest assured that we are doing our best to make the network better each day. Have a look on what we are working on by visiting the roadmap page.

You MUST HAVE some kind of warning system before the servers resets AND/OR some way to not make them reset at all. How do you think the users feel when we put tens of hours or days into initial configs and then everything is just wiped out. Rhetorical question so I'll skip the question mark.

We have planned to implement events in the network. Although for most maintenance work you would receive an event prior to maintenance, sudden downtime and resets is a reality as the network is permissionless and has no fine-grained control over the providers. Check AEP-59.

The platform itself MUST HAVE built in protection from resets/wipe-outs. Possibly some kind of redundancy. Otherwise it's just completely unreliable. I mean users will stay away from the product. The worst thing that can happen as a server provider is allowing the customer's servers to reset. This MUST NOT HAPPEN. I understand the value of your token is growing. But THIS SHOULD NOT BE THE MAIN POINT. The main point should be to provide a reliable server service. As I said there is no warning, no explanation, no replies to issues, nothing.

Related to this there are several AEPs, such as AEP-49 and AEP-52. Let me know if you'd like more context. Note that those are still being designed and it's not guaranteed that a solution about moving data from one provider to another will be supported.

Also perhaps reply to issues here within days instead of within weeks or not at all. If you have a link for the github discussions please post it here.

https://github.com/orgs/akash-network/discussions

You can also hire me for a while as a kind of tester or something because I actually like Akash, theoretically, if it would actually work (for me). And I'm not expensive. I understand the feelings of the users. There are several apps in the Android play store that have functions which I have suggested, as a kind of resume.

There are several ways you can get involved and get paid for it. Check the funding programs.

Have you considered teaming up with Yggdrasil somehow?

Not that I know of. Will have a read about that project!

Hope I helped clarify! 😄

@bitcoinmeetups
Copy link
Author

I wish there were more competition in this space. You would benefit from more competition. You are almost the only service that provides - kind of - decentralized servers.

I also feel like there should be a place for users who develop stuff more sprawlingly. Running any of the services you listed does not appeal to me because I can already predict what it would be like. So there is not so much of a point to do it.

Look I am a fun and friendly person. I don't feel like I'm asking for much. Just a tiny, reliable decentralized server with small storage and freedom to apt around. Please cater to me also, not just to the bigger companies and startups only.

@bitcoinmeetups
Copy link
Author

Is there any third party service building on top of Akash that could cater to programmers like myself?

Akash > Third party with reliable and free as in freedom SSH'able servers > Myself.

That could work, if you don't want to do it yourself.

@bitcoinmeetups
Copy link
Author

bitcoinmeetups commented Feb 3, 2025

I make so many changes to my servers so I can't use a pre-prepared container because there are constant updates and changes. And if it's unreliable then it doesn't work at all.

Perhaps as a workaround you know of one specific provider to Akash who has more reliable servers than the others that you could recommend? And perhaps some to avoid so I don't fall in the same trap over and over.

How about this:

You could make a list of which servers are more reliable.

So users will simply not choose server providers with servers that reset.

And/or reward providers for reliability/redundancy to incentivize.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants