Update: I have put out a call for FlexPool to open up the source code to FlexFarmer.
FlexPool has just released a public beta of their own lightweight farming software to enable Chia farming with their pool without running a local full_node and going through all the sync issues. You can try it out here. Right now it is only available for Linux, either to install or deploy as a docker. Windows is coming.
This is a lot more reminiscent of modern Bitcoin and Ethereum mining software where the pools runs the node and the farmers just do lookups. And they are also promising better performance there. The full list of features they are providing is:
- No node sync required
- No local wallet required (you will need to input a wallet address)
- Fast lookup times in milliseconds rather than seconds reducing stales to almost 0
- Better identification of plots and possible issues
- Naming of each individual rig for website dashboard stat tracking
- Better tracking of stats and stales/invalids
- Low processor usage
- Low RAM usage
- Low power usage
- Low memory/storage space usage
- Much more stable
- Reports plot size to pool (similar to reported hashrate in crypto mining)
- Autodetects new plots added
- Failover to second configured region during server outage
This really sounds like it is going to radically reduce the overhead to farming Chia, but because of the nature of what they are running here (the back-end infrastructure will farm the blocks, not the farmers) it will only be usable on Flexpool unless they release the back-end and other pools implement it. It also goes against the nature of decentralization that Bram and company envisioned. That said, I agree with the direction. Requiring everyone that wants to participate in the Chia blockchain to run a full_node and full wallet is simply untenable and the ecosystem will not grow under those conditions.
The tool looks a LOT like traditional crypto mining software, and is as lean and performant as you would expect. It takes up no space, gets setup instantly and starts farming your plots right away. With FlexPool. This is an important step forward, and seems to do lookups at least 10% faster than my full fat Chia client. I was already getting good lookup times so this is very good and should fix problems for people that have slow lookups.
The version I am testing is Alpha 0.8 on Windows and it feels release ready to me. You need to use the extract_farmer_key.py python script to put your private key into the program in order to read your plots. This doesn’t appear to be doing anything but display it on the screen but I am not a programmer. This portion is, however, just open-source and not closed at all and feels fine to me. If you are going to be nervous about any of this, this is the step. Its not really possible to read your plots without it though, so there is that.
One of the features they have been discussing is that it should run just fine on a a RaspberryPi 4 and I think they are correct. The thing is extremely lightweight. This is making me reconsider the full PC farmer entirely, and possibly just run the whole setup off a docker without any fuss. It also should run just fine as a service on Windows if you want that functionality.
While we were testing this, the famous Mad Max actually seems to have won the very first block farming with the new FlexFarmer. I thought that was pretty cool. I think that means its likely he plotted the first plot with non-Chia code and farmed the first block with non-Chia code. Go Max!
I am incredibly excited about this, as well as the upcoming Nucle wallet that has been announced. Between these two products Chia farming just got a whole lot more accessible. If FlexPool decides to release this setup to the community so that people can run their own back ends and other pools can incorporate this I think they will have won my “contest”.
Why are you so against everything that Bram and the company are saying lately? You make the point yourself in the article that it goes against the vision they had about decentralisation. I respect that very much and I can’t believe they are actually sticking to that. Your argument that it will slow growth is irrelevant because the network is already at record nodes count. I believe they are doing the right thing by changing the focus more on what the purpose of a cryptocurrency is rather than keep improving the convenience of pool operators and farmers. I believe the network is big enough for maintaining the security of the blockchain, so it is more imperative now to make it more accessible to the general public through exchanges and wallet tools that will allow using the coins without being a farmer with full node setup.
I think that as long as the Chia client most of those nodes are running is a python shell with inbound open ports into it that the Chia community and blockchain is a single Python RCE (4 since 2014) from being totally taken out.
Running a Full Node is just a miserable experience. Even in Docker mine is using 37% of the CPU and 10GB of ram. Syncing can be a weeklong experience unless we’re talking about the miserable Wallet, then it’s a months long experience. And nothing seems to corrupt easier than that wallet DB. A viable, stable implementation is important … no matter all of the theoretical benefits. Chia has to be able to overcome crappy software if they want to be taken seriously. I mean, how does a third party come along and write a plotter that absolutely destroys the formal Chia implementation … it’s like no one at Chia has ever heard of multi-threading.