My misunderstanding is because I never grasped the difference between farming and harvesting.
If I set up a new PC, no plotting, just has drives with plots installed on it, then would a farmer or harvester need to be installed on it to feed results back to my Full Node PC?
Take a look at your Connections tab on Full node panel. You will see there are two local connections (127.0.0.1): your farmer and your wallet. On the other hand, if you go to Farm panel, and expand your āadvancedā section, you will see just harvesters there (no farmers).
What does that breakdown mean? That the Full Node is for the farmer, but any other PCs will run only the harvester? That the harvesters report to the farmer, and there is only one farmer, and he is on the Full Node?
My understanding is that the full node is the process that deals with blockchain. Farmer, as you set gets data from all harvesters, and potentially reports to your pool. Harvesters are just dealing with your HDs and only interact with your farmer.
The main problen now of your project is the ā2 gb DDR2ā. I think tās necessary at least 4gb of ram to farm chia.
And sure, Iām talking about running chia in a clear linux distribution like Debian. I donāt recommend you install windows 10 on this low spec hardware. Have you read the microsoft system requirement to run windows 10? I can remember they saw the need of have at least 2gb to run the system. But if you add appāsā¦
Soā¦ to answer the question, yes theoretically, if the wind blows the correct way, it could workā¦ but not reliably. I spent hours yesterday trying to get everything perfect, but my patience just wasnāt having it. I got it to sync up, only to realize that I didnāt add directories. Went to add directories, and it all went to heckā¦ I turned it off and sat it in a corner lol
Full node - this syncs the blockchain with peers - nothing more. Does not require your private key. Only run one of these.
Farmer - this listens for challenges from the full node and forwards them to all connected harvesters. If a harvester responds with a proof, the farmer signs the block with your private key and forwards to full node for inclusion into blockchain. Needs your private key. Only run one of these.
Harvester - receives challenges from farmer and checks plots. Sends proofs back to farmer if found. Does not need your private key. Run as many of these as needed.
Wallet - syncs your transactions from full node. Needs your private key. Only run one of these.
All of these can be run on the same machine (default installation) OR you could run them all on separate machines.
That was the best and clearest explanation I have read.
As to:
Please define āpeersā. With whom is the full node syncing?
Do you mean that the full node is what is kept in sync, by way of the harvesters (peers?) feeding the farmers who feed the full node which feeds the blockchain and stays synced?
That is what my next endeavor will be; to split off the farming.
And if I have it correct, my full node will run everything, and my other PCs (just sitting on plots) will run only farmers?
In this case I mean other full nodes out there on the internet. All those connections you see at the bottom of your āFull Nodeā tab are your other full node peers out there on the internet. This full node network is responsible for keeping the blockchain database itself up-to-date across the network on all machines. The farmer/harvester/wallet processes make use of the full nodeās database to do their work.
Almost - your āmain PCā will run your full node, farmer and probably wallet. You could put each of these on their own machine, but thatās probably overkill. Then you can have as many harvesters as you need, each running on their own PC/raspberry PI, each āharvestingā a bunch of plots. In this way, you can scale up to tens of thousands of plots overall.
Difference between āfarmerā and āharvesterā:
Think of the farmer as the āownerā of your farm. He listens for challenges from the full node and yells them out to his workers in the field (the harvesters). Those harvesters all run around the field and look for proofs. If one of them finds one, it yells back to the farmer. The farmer verifies it, signs it, and sends it on to market (the full node).
Do you still have that box around? Could you modify your config.yaml to limit the number of peers down to 10? My understanding (maybe wrong) is that most of the time used by chia is dealing with peers.
In the āfull_nodeā section change this line (from 80 to 10):
target_peer_count: 10
Also, adding a $10 PCI/NVMe card to hold blockchain/wallet dbs could help.
Yes, sure Win will be a hog on such box, as UI will be the main CPU/RAM consumer. However, it may be that chia (when left alone) may still run nicely.