Hosting blockchain services on Threefold

I understand that literally anything can be hosted on TF, but the demographic that will innately be most interested in using our services are other blockchain projects. Specifically, I am referring to nodes and staking. These services require physical hardware to run and it is often done on a 3rd party host. Staking nodes need very high uptime to avoid slashing so many avoid hosting self-hosting these. The upcoming gold certified tier is perfect for this.

So how can TF get in on this? IM GLAD YOU ASKED

Two ways really. TF can deploy their own marketplace to launch nodes on the grid OR a partner can launch this marketplace with TF as the backbone.

How hard is this to do?

It does not need to be built from scratch. The biggest player in this field is allnodes.com, but they are not open source. However, Ankr is open source (edit: I think, not 100% sure). Check out these links.

Just like New Orleans runs on beads, this marketplace needs to be configured to run on TFT.

PRIME opportunity here, but unless you want a blinking LED on a breadboard, I’m not your guy to set this up. If anyone wants to form a team though, I’m all ears. It would be quite the undertaking. The commissions of such an endeavor could be enormous.

PS also see blockdaemon

8 Likes

This would be an incredible multiplier for the usage of the grid and from my understanding TF fits perfectly for this. So glad you bring this up. We should address this asap. I mean this already happens since it’s possible f.e. to run presearch nodes on TF grid but providing resources for a wide spreaded node housing/staking platform would be high level! I’m afraid I wont be any help in developing this but I would be happy to provide access to high-available 3nodes running on powerful servers under DC conditions.

Count me in here. But something like a solana node, or even more intensive a solana RPC node, requires some serious hardware and can’t just be distributed amongst a bunch of servers that suitable for “lighter” workloads.

Maybe different nodes and their different hardware requirement could almost me black box offerings, you just click on one and it spins up on certified, benchmarked hardware that will meet that projects criteria. And ETH node or Cronos node would be less intensive, down to the basic Flux node and the two new micro versions, and then just your basic master node. A third party, cough cough, could provide consulting services or guidance to the clients, while opening it up a bit to several qualified providers as some node require hardware that costs a major chunk. I’ve run many nodes, validators, master nodes, etc and was working on the side in spinning up a solona rpc and a virtualization platform i the same rack at an Equinix data center for NFT botters and the like.

So I love where you head is at, there are just some defiance logistical issue to get through, but I think would be worth it.

Horrible typos above, but I might need to spin up a Geth (ETH) node and an L2 node with it…if I get frisky, I might try it on threefold, despite having the hardware myself…will keep thread updated.

HI @FLnelson. I spoke with Ankr’s founder/CEO (https://www.linkedin.com/in/yunfansong/) a long time ago, exactly with the thought you posted here in the forum. We went back and forth for a while, and then he decided that he was too busy on boarding new customers and deploying bare metal servers in colo around the world.

I take your post as an indicator to revisit that discussion. :pray: you!

Also - there is a group of people that are looking to have presearch on the TF Grid, and another group that is looking towards (name escapes me at this point in time…)

3 Likes

Are you thinking about Hydra?

2 Likes

Yes! Thank you for refreshing my mind.

Then this lull in the market should be a perfect time. It’s the perfect time to strike deals.

1 Like

A community member has already started on this project, its a proof of concept mostly at the moment, but check it out zonaris.com

At the very least I think it’s a good name.

1 Like

Sent in the form about staking ankr to run your own node you can apparently rent out spare capacity on. Wasn’t sure what kind of hardware they would be using, as needs can vary vastly from chain to chain. This is really intriguing to me though, as I run some nodes for myself and am dipping my toe in providing some node services. While also providing threefold capacity. Mixing the two sounds like fun! I can’t see what I am typing as the cookie acceptance won’t go away, no matter how many times I click. So I hope this is readable without a lot of typos.

4 Likes

You have done a stirling job, no typos seen while I was reading. Mixing things up is always good: spread risk, discovering new flavor combination to name a few!

No response yet from the Ankr CEO, will reach out again and see if he responds.

2 Likes

For some reason this was within my view, which it sometimes takes to get my attention, as I responded to another post and I remembered to follow up. I spun up a threefold instance twice, which took forever and two weeks each time, but could never even ping the static ip’s that were supposedly assigned to them, so I didn’t get farther in my experiment of running a node for, I think Arbitrum, is what I was going to try.

Also, the filters to find a unit to rent on didn’t function, so I had to go to the explorer and grab a farm ID of somebody with enough resources and put that it to create my instances. And the explorer and I have since apparently broken up and are no longer on speaking terms. At least it thinks so….lol.

1 Like

Thanks @LongStreetTech for this feedback.

That’s the kind of user experience we want to know about in order to update and improve the TF Grid.

There are planned upgraded and improvements on the way for the filters and stats shown by 3nodes/farms for users to pick the proper workload.

Did you run a docker of Arbitrum via a Virtual Machine on TF play.grid.tf or you converted the docker into an flist on https://hub.grid.tf/?

Hopefully next time you try it, it will be easier and faster.

@Mik, I used play.grid to deploy an ubuntu official flist. I know Arbitrum wants you to run a docker image, but I could never get the chain data to persist, despite using correct syntax/mapping. So I just wanted a vanilla Ubuntu VM to install directly on to.

So here is some positive news. The explorer and I are apparently on speaking terms again and it is no longer taking an eternity or not working at all. Maybe that was just a temporary thing or somehow related to the machine I was using. Previously I had used an ssh key pair created with puttygen and then tried to use putty to ssh in. Just now, I spun up another the same way with default Ubuntu flist, but from a native Linux machine which I generated a key pair for my user. The machine spun up in seconds, or more like a minute, versus the literal hour(s) on my previous two attempts. So huge turn off now averted. Again, maybe it was me, but I was using the web ui, so I don’t know what I could have done to have caused the tremendously long deploy times. However, on this new instance, I still cannot ping the public ip. And on key exchange when trying to ssh in from the Linux terminal, I get a terminated by peer error. Also, seriously 32 thread and 256gb memory limits? Lol. That’s less than the machine I browse the web on….well, actually was having some issue so took a few dimms out and currently 128GB of ddr4-3600. Then again, I’m a fool. Most people don’t need a threadripper for web browsing. :wink:

Also, the issue may be farmer based…possibly, as there is only one US based farm ID that has met the specs I was trying to get. Other say 1 public IP available, but then fail for lack of resource when you try to deploy.

I’m going to give this instance a couple more minutes in case it is still initializing, then destroy and maybe try a small instance from another farmer and see if I can at least access it. Honestly, for me personally, a lot of the validator or rpc nodes are obviously more efficient to deploy on my own bare metal, but as I am looking for innovative ways to get capacity to build a service on, other than just buying more servers, wanted to try this out. The issue is hardware performance. I need fast solana RPC nodes. They have reasonably stiff hardware requirements. Right now my dual 2697v3 w/512gb ddr4 and a gpu and nvme storage is in a battle. My threefold capacity I provide is based on the x20 series of poweredge servers, so v2 xeons and ddr3. Run a lot of personal workloads on that kind of hardware still and think very viable for use on the threefold grid, but there is a market for those farmers that have newer hardware. I’m not going to say more expensive, bc my racks weren’t cheap back when I got them. Lol.

Anyway, I digressed, but will keep you posted. I might also write a docker image for something and use terraform to try and deploy it. New to terraform but played with it a little and looks easy enough, at least for basics. Just have limited time to play.

Thanks for the positive reinforcement!

-Robert

1 Like

@LongStreetTech

Not being able to ping the ip indicates they do not have their public ip set up correctly. I would try another farm to experiment on.

On which node id did you deploy?

1 Like

I guess that’s @ParkerS farm. As far as I know pub IP settings where checked by waynand and should be working.

You can try to deploy on node 1983. this one has a fixed public IP and subdomain assigned. Or you can go with node 1895. this is configured and tested and has a “rentable” public IP available

1 Like

Can you tell me if you attempted anything on 2914 or 3049, of these boxes are having issues then I will get them fixed just need to know a problem exsists as far as I’ve been informed they should be fully functional. You can also let me know what you need on the net from a U.S. host and I can try to source it for you. There will be a large upgrade cycle early this week on the farm that includes hardware and additional ips. Hopefully we can get you squared away!

1 Like

Thanks for all the great replies guys! The first one was, going from memory as I am at my parents now, something like farm 214, it has quite a few nodes with chunks of public IP’s.

I need to figure out how to assign public IP’s at some point, as I am deploying some units whe they are fully upgraded to a location I am have several blocks.

I think my final test on a “small” instance was a node in the 19xx range. I’ll see if I can see a deployment history later.

@ParkerS if you want to drop me a DM maybe we can talk hardware and keep it out of this thread. I’m not sure it is economical, depending on use case, vs bare metal I own, but worth investigating. Especially for edge or geolocation access.

Thanks again for all the responses. I’m not saying ankr is going to have a competitor, lol, or I’ll even have any luck, but did want to look at threefold as a node deployment option. I like the project that, I think, was linked early in this thread. Maybe I’ll try to reach out to them.

Thanks again!

-Robert

1 Like

@sigzag, it won’t let me deploy using even a portion of your ssd space, claiming there isn’t enough available. I was trying 2tb, but scaled it down, split it into drives, etc and same error. Tried both id’s.

@ParkerS, got no PM, but the two nodes you mentioned were lacking in resources. May try to deploy something basic to at least test my connectivity. Don’t go out of your way for me, I can deploy my own bare metal. Either on threefold to rent from myself, which seems silly, or for use case. Problem is these stupid new machines and expensive ass ddr4, when you need it in the 1/2 and 1 terabyte range. Sold a half terabyte a little while back and wish I handn’t.

Anyway, appreciate all the feedback guys!