So I’ve been farming since April 12th and up to 340 plots with 0 XCH rewards and trying to work out if I’ve just been unlucky or there is something wrong with my setup. Looking through all the github issues it seems other are running into issues, potentially with running a remote harvester connected to a full node (example).
My setup up until today has been a main PC running the full node (with the GUI) and also plotting, and 3x plotting machines running only the harvester (pretty much following the guide on Farming on many machines). It’s somewhat of a temporary setup until I finish building a proper NAS server with all the plotters also having multiple hdd’s attached. Everything appeared to be working correctly in that setup (ports open, remote harvesters accepting and responding to challenges in < 500ms, no errors in logs on any machine, chia plots check sees no issues, pool and farmer reward address set properly in GUI and also reflected in the config file, target address shows nothing on chiaexplorer etc). Doing a grep on my logfiles I’m unable to find any proofs, though I just have full logs going back 3-4 days.
So today I switched things up and each remote plotter no longer runs the remote harvester and the main PC running the full node just accesses the plots over the network (over gvfs sftp), and again looking at the logs everything looks fine and it’s attempting proofs on all plots in < 500ms. So hoping I’ve just been unlucky to date and there aren’t other issues, or if there is some bug with remote harvesting hopefully the new setup won’t have the same issues. Anyone else experiencing the same sort of issues?
I’ve had no luck running remote harvesters. I’m switching back to individual nodes for now, consolidating a few plot storage locations on my network. Remote harvesters are a giant pain to set up and I can’t figure out how to monitor them. Are there bugs in it? Probably, and I have no good way to tell.
I’m having trouble understanding what the downside is to just running multiple full nodes? Remote harvester seems more complex than it needs to be, but I might be missing something.
I tend to agree; the only downside of a full node is that it’s resource intensive (relatively speaking) and you have to make sure it doesn’t fight with the main node for port 8444 traffic … but if you have upnp disabled in the chia config, and have manually forwarded the port on your router to your “main” farmer, that seems safe enough to me?
I got 2 wins on a remote harvester. I think as long as you can see plots passing the filter on the full node (judged by the plot count in the GUI farm screen) you’re ok.
It’s tough though. I only have 3 wins total and have moved from a temporary plotting setup to a permanent farming setup since my last win. It’s been 7 days since I won with 4 days estimated. I was just barely keeping up with the netspace. Now that I’m done plotting I’ve already hit a 5 days to win estimate and I’m anxiously awaiting a win so I know my setup is working.
I just completed setting up a remote harvester on 1 of my 2 NAS, using an Ubuntu docker image…
In the harvester logs, and on the main node, I see that it’s sending new_signage_point data to the remote harvester, and farming_info is going back to the main node… I see “x plots eligible” in the remote harvester logs.
What I don’t see is whether the main node is truly using that data… The “total z plots” count doesn’t include them. (now that I removed the mounted path which is now covered by the remote harvester)
Going to finish setting up a remote harvester on NAS #2, just curious what the logs are going to say when the main node has no actual plots, only 2 remote harvesters.
It’s a QNAP 8 bay NAS with a quad-core ARM chip… The docker image provided by chia-network doesn’t support ARM, so I had to start with a Ubuntu image (bionic) and build up. Copied the CA and generated the init properly, added my keys, edited the config, etc.
It has my giant plots folder mounted (The QNAP container station app supports picking folder(s) and giving them mount points inside the container).
The remote harvester logs look good… I hope they update the main node to reflect this so I won’t be constantly checking the remotes.
I’m really surprised no one has taken you up on this. I mean, they are NAS owned by Jeff freaking Atwood. If you’ll sign them with a sharpie marker, I might even buy them myself.
They will be updating it. It’s an issue they’re aware of in Github. Scared the crap out of me at first when I didn’t see my harvester plots accounted for in the GUI.
Finished the second remote harvester, and shut down the harvester on my full node. Looks like it’s working very well. Log sample:
Full node:
2021-05-03T06:33:45.968 farmer farmer_server : INFO -> new_signage_point_harvester to peer 192.168.141.104 adcxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
2021-05-03T06:33:45.969 farmer farmer_server : INFO -> new_signage_point_harvester to peer 192.168.141.123 dccxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
2021-05-03T06:33:46.002 farmer farmer_server : INFO <- farming_info from peer adcxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx 192.168.141.104
2021-05-03T06:33:46.021 farmer farmer_server : INFO <- farming_info from peer dccxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx 192.168.141.123
Harvester1:
2021-05-03T06:35:50.667 harvester harvester_server : INFO <- new_signage_point_harvester from peer f6bxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx 192.168.141.129
2021-05-03T06:35:51.066 harvester chia.harvester.harvester: INFO 2 plots were eligible for farming d709735168... Found 0 proofs. Time: 0.39780 s. Total 260 plots
2021-05-03T06:35:51.067 harvester harvester_server : INFO -> farming_info to peer 192.168.141.129 f6bxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
Harvester2:
2021-05-03T06:36:07.763 harvester harvester_server : INFO <- new_signage_point_harvester from peer f6bxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx 192.168.141.129
2021-05-03T06:36:07.943 harvester chia.harvester.harvester: INFO 1 plots were eligible for farming d709735168... Found 0 proofs. Time: 0.17893 s. Total 350 plots
2021-05-03T06:36:07.944 harvester harvester_server : INFO -> farming_info to peer 192.168.141.129 f6bxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
One thing I’m excited about with this is that I am pretty sure I can plot again on the full node. I had just switched to this node for the fast response times being on the same switch, but whenever it would copy a plot to the NAS I’d have 15 minutes of warnings. Started the plotter script, so I’ll find out tomorrow, but I’m optimistic.
Update after 24 hours of this… Remote harvesters on NAS is awesome.
Response times are consistently less than 0.5 seconds and I haven’t seen a single response even hit 2 seconds. I don’t need to worry about network congestion, and copying files from USB doesn’t affect speed at all.
I bought codinghorror’s remaining NAS, and I’m going to shuck a bunch of drives and set them all up, within a week or so I should have 5 remote harvesters and 8 or 9 plotters. Still wish I could plot faster, but I really don’t have the time to hyper-optimize that. (at 661 plots now)
Yes. A vanilla Ubuntu 18.04 docker image that I then configured. Did this on 2 separate NAS and now my full node/farmer has no plots itself, just has the 2 remote harvesters.
Would like to hear this too… I’ve been running with only remote harvesters for a week, no blocks yet (but my estimate to win with 1400 plots is 8 days).
BRO! You’ve more than doubled your plots in 12 days. Nice job! Are you seeing regular rewards? I’ve been dry for going on 3 weeks. This network growth is
awesome.
Are you seeing regular rewards? I’ve been dry for going on 3 weeks. This network growth is 