Cannot Get Farmer to Sync - HELP!

I deleted all the plots, went through all the steps again setting up MM. I’m recreating the C8 plots again now. Trying everything I can think of, but still no luck with getting Chia to Sync on this Proxmox VM. If anyone has more ideas, I’m more than happy to keep trying things…

Thank you,

I would really look into how proxmox is handling network - Network Configuration - Proxmox VE

Looks like the default setup should be a bridged mode, but maybe in your setup it is not, or some extra settings need to be specified. I don’t think that the error coming from openssl / s_client has anything to do with MM or Chia setup.

Actually, maybe it would be worthwhile to install chia first, and see whether it will get some peers. If it will have the same problems, that could potentially eliminate MM as the source of those problems.

Lastly, your plots are connected to your mnemonics, so when you do a clean reinstall, just use the old keys. There is no reason to kill those plots.

I could create a new VM & just install Chia on that one. I’ll do that this afternoon.

I deleted the plots just in case something wasn’t 100% correct with them when I last created them. Re-doing things is making me faster / better with all this stuff. Plus the GPU plotting on an RTX 3060 is just over 4 minutes, so its a pretty fast process. I might delete all the plots & redo them just to see how much faster an RTX 3070 is or RTX 3060Ti is just for fun. This is the current network device settings:

Thank you,

Everything on that screenshot looks good to me, but again I have never worked with proxmox, so whatever I think about it may mean not much. Although, maybe that Model section is the problem. Have you tried that Intel E10000 or Realtek RTL8139? Also, could you expand that Bridge drop-down?

If you spin another VM and install Chia, just run that openssl / s_client to see whether you get that error. It is like grasping for straws, but the best guess on my side (that proxmox is somehow messing it up or MM farmer cannot handle such setup).

By the way, to test whether that SSL is really the culprit, you don’t need to have any plots. Once you start the node (even without any plots), it will immediately try to create both dbs (bc and wallet if not there), and in order to do that it needs to connect to some peers (at least 1). As long as those connection errors are there, that should be the main focus right now. The wallet db is really fast to build (sub-hour, I think), so you can blow away that at will. The bc takes longer to sync, but my understanding is that once it will start working, you will be able to download bc db directly from Chia to speed up the syncing process (don’t know why they are not providing monthly snapshots). Also, just for testing, you don’t need to create new mnemonics. You can just start the GUI, and it will go right away to that process, as bc db is independent from your mnemonics (thus can be shared).

from reading all this it sounds like its time to just start over… u took a wrong hard left somewhere. my farms down right now els i would try and replicate. an h200 died on me should be here in a couple days if your still having trouble i will actually help…

questions tho.
why did you decide to run Ubuntu >server< in a vm? u can spin up chia in a lxc container in a fraction of the time with a template and have your shell with direct access to hardware resources… that alone could fix your issue.
easy enough to test in 10 minutes with proxmox.
personally i run proxmox with a ubuntu desktop vm minimal (gui) install. with just only chia installed. and containers for a harvester. for many reasons i chose this that im not gonna go into.
but something seems off
the main goal in proxmox is to keep prxmox as clean as possible and to keep each process happening somewhere els. isolated, safe, repeatable. example: if u need to mount a disk, create a container specifically to mount that one disk than share that disk from that container to the rest of the vms and containers. treat every proccess like this.
i say this because i know nothing of gigahorse or any of the forks or any of that. but thats how i would approach it in proxmox. so you would be able to know wich part of the whole machine is failing.

it puzzles me because a major perk to running a vm in proxmox is that you get the gui… but you chose no gui… when your admittedly new… if your taking this seriously you should of started with Ubuntu desktop…
before i help at all tho
always take a snapshot of your instance before changing something and testing. so you can always go back… and again try to not change anything on proxmox whatsoever.

if i where in your exact shoes tho facing the same issue.

id try in software some simple tests. in this order pinging, port knocking, finding the route in my router, making sure everything was open, disabling firewalls, contacting isp, configuring networking in proxmox(wich must be done)… both ways from in proxmox and inside the vm. test in a container the config.

if theirs still issue than id install my favorite admin panel. such as cockpit. and take a good look at everything inside the vm.
if still issue id use history command to see everything id done and find the miss step manually line by line.

if i failed at that

than id for sure blame the creators of all the stupid forks. for adding a lvl of difficulty lliterally NO ONE was asking for. and go straight back to using OG chia client pure. with just mad max for plot creation in a separate container. and another container for the harvester.

sorry i cant be of any actual help right now.
the best part of proxmox is that you can start over what your doing very very easily. using templates, history commands, snapshots, replication, high availability, backups. all for the purpose of making problems llike your facing trivial. take advantage of proxmoxs full feature set.

if i think of anything els smart to say ill update when my part gets here than ill help

how have u setup ur proxmox networking? do u have other machines in proxmox that are able to use internet ect. .

Hello jonesjr, thank you for this long reply!

I’ve already re-done the plotter & farmer VM’s several times, one to learn & get better / faster at setting this stuff up, two I enjoy this. I don’t think completely starting over would be the smartest thing, otherwise I could end up right back here & not know why. Or I could end up re-doing everything completely & find out it wasn’t even something in the installation / setup.

I was starting to get into Chia with 18TB external drives, but then I came across Digital Spaceports videos on YT. So I followed what he figured out, instead of re-inventing a new setup & bought a Dell R720XD to get started with. I followed his setup in Proxmox & what he did & that’s how I am here. I did not have any experience with Proxmox before this, only Hyper V, which is fine for my Internet marketing business. I don’t do IT for a living, I just keep my office network running. I have Ubuntu Desktop running on the Farmer VM. Ubuntu Server I think gets you faster plot times, so why not… Speed & compressed plots was what got me interested in this setup in the beginning.

“lxc container in a fraction of the time with a template and have your shell with direct access to hardware resources… that alone could fix your issue.” I have no experience with that yet. I tried setting up a Windows 11 VM, but the Gigahorse Plotter just would not run. It would start, but right at the last step, it would just sit there & not start creating plots. I saw several people talk about how much faster Linux was at creating plots so I switched my setup & got into Proxmox.

“it puzzles me because a major perk to running a vm in proxmox is that you get the gui… but you chose no gui… when your admittedly new… if your taking this seriously you should of started with Ubuntu desktop…” Again, I was just following Digital Spaceports setup (He even advises against using the GUI / Desktop for some reason), which I’m not at all attached to his setup or what I currently have. I’m completely open to redoing the Plotter or Farmer VM’s in another way. The TrueNAS VM is working really well, so that one I want to keep it as it is.

I was taking backups at key points, before proceeding & making more changes. I ran into two issues so far, where even after restoring the backup, the error persisted (Which should have been fixed I thought). I just deleted the VM & did it again. It’s one way to get better at doing this, so it didn’t bother me too much. Although I have been spending a lot of time with this & I need to get back to focusing on clients… At this point I just want to find a setup that works, creates the compressed plots & starts to farm them. I will try taking snapshots now insteaf of backups though, again this is totally new to me in Proxmox. In Hyper V, I jsut shut the VM down, disconnect the Hard Drive, copy it somewhere for a backup & then start it back up. I know Hyper V is limited, but its been perfect for what I’ve needed in my office network & keeping that more simple is a main goal.

“id try in software some simple tests. in this order pinging, port knocking, finding the route in my router, making sure everything was open, disabling firewalls, contacting isp, configuring networking in proxmox(wich must be done)… both ways from in proxmox and inside the vm. test in a container the config.” This is where I’m up to now. One of my clients is an IT company & they assist me when things are out of my scope with the network stuff. I am not familar with “Cockpit” but I’m open to learning more new things to figure this out.

I have the OG Chia client runnning on another computer, creating plots the turtle way, even with a Threadripper 3960 & blazing fast SSD & ram.

“the best part of proxmox is that you can start over what your doing very very easily. using templates, history commands, snapshots, replication, high availability, backups. all for the purpose of making problems llike your facing trivial. take advantage of proxmoxs full feature set.” I am beginning to get familiar with doing this like this & developing ways to make myself more efficient.

I’ve just used the default network settings, so far every VM I’ve setup has been able to reach the Internet.

Gigahorse allows you to create compressed plots, which with C8 compression are around 71GB. So I can get a lot more plots onto the same drive space than uncompressed plots. The normal Chia client can’t run compressed plots yet, so you have to use the Gigahorse farmer if you’re going to go with compressed plots…

I will go through what you wrote more though & begin thinking about how I can adjust my setup better & hopefully get these compressed plots working soon.

Thank you!

-never to late to retry. proxmox backup server is amazing if you where wondering.
-a vm will never beat a container as far as preformance is concerned. its imposible. its a fact.
core for core… a vm doesnt have direct access to hardware, at all. aslso make sure your setting your cpu correctly when setting up the vm. with correct kernel. im hoping you have read how exactly to install windows from the proxmox docs…
-with vms theirs much overhead. a server with 24 cores. can do say 10vms… or 50 containers. always use containers if u can. ull get alot more out of your system. privileged…

  • but if you’ve setup windows correctly with the proxmoxs special drivers for a windows vm… if not followed correctly can cause alot of issues for a windows vm. programs will refuse to work. gpu wont utilize, vnic gets messed up lots of strange issues.

–why not just use mad maxs cpu plotter… i had a plot every 20 minutes plotting to ram in my t710.
plenty fast for most humans… and extremely effective when setup in a Ubuntu lxc container. just update your templets in proxmox. choose ubuntu server… than create ct and ur done. save that ct as a template after you apt update. so u always have a fresh new one ready to go… (also linked clones are cool af explore that…) than its a simple install for madmax. its just a couple lines of code and is actually endorsed and recognized by chia as safe and working with OG client… than setup another container to mount all your plot hard drives to.(u can mount the same hard drive to 2 different containers at the same time…) so mad max can feed plotts to hdd while your harvester harvests them… seperately… at the same time… and harvester pointed at your already synced OG client u have. and your done. farm completed.

20 minuts. uv got everything u need to do this.

not sure how many things you have running on proxmox server but if its more than 3 you deffffffinently want to configure your networking properly.… if ur using default… than ur only using one network port huh?.. im not gonna go into why. but you need to setup bonding, and give that connection to your vms… and give a dedicated eth0 to your proxmox host ip address…
network traffic gets outa hand extremely fast in proxmox if not properly mitigated. especially if u have a bunch of vms. will cause lots of communication erros. slow host. ul loose ur host pve gui if u dont give it its own dedicated eth1… and it will cause u days of pain. ssooo
plug an ethernet cable into as many ports as you can. back to a router
setup bonding for all thos(eth1 eth2 eth3) connections except for 1.(eth0) give that last 1 whole physical connection to your proxmox host. want to keep management traffic separate. segregate everything… virtualize everything… isolate even the smallest proccess in a lxc container.
you must accomplish this. or there will be problems.intermittent connections. chia wont stay synced. lots of issues for sure.
if not done correctly while attempting u may loose your gui for a bit.
networking’s a major part of how proxmox works… they recommend 10 gig nic minimum to take advantage of full feature sets… live migration, HA ect. its really wild shit to see the first time… iv ben in a minecraft server vm… whilst that vm was being moved to another physical server… didnt even notice on the frontend…

ohy ya templets require zfs to get all features. so make sure u have a pool setup

gigahorses plots are not very usefull if you cant get the stup working… who knows could all be propaganda to slow you down, make u loose focus… question everything. or you will get taken for a ride.

good luck to you sr. leme know if u have questions or u get stuck. i can be more specific… but the keywords i use should help you get to the right place.

–tell me why you want to plot so fast with such a small amount of storage? where ya going?/
u should only follow chias instuructions… but thats another conversation…

I’m really happy with TrueNAS right now, but once I get this working, maybe I will take a look at Proxmox backup.

I was just used to using VM’s in Hyper V, but I’ve heard about containers or linked clones & how much better they are. I am going to find some YouTube videos about conttainers after I send this.

I will also look at the Proxmox docs for Windows too after this.

I have the TrueNAS vm, Plotter Vm, Farmer VM, an Ubuntu desktop VM for testing that is usually off, a Windows 10 VM & Windows 11 VM both just for testing & are usually off. I’m going to look at using Containers instead though after this message. I’m also going to look at Proxmoz networking too. I am using just the eth0. The physical server has its own IP through its own physical NIC, the Gateway haas its own IP too. Then each VM has its own IP. I have two more physical NICs I could use in the server. The GUI is accessed through its own dedicated IP. Everythign else uses its own IP given to it from DHCP, but I will use the other two NICs & be sure to keep the GUI on its own dedicated IP like you said.

I have a Mellanox IS5022 InfiniBand InfiniScale 8-Port 40Gb/s Networking Switch, but I have not gotten up to adding that to my network yet. Once I get this MM Chia Plotter working & farming, I will work on getting that into the network. I was going to add that when I got to adding a JBOD to the network.

I am using ZFS currently.

I was getting 3.5 minute plots with the 3060 using the MadMax GPU plotter. I used to race MotoCross, so the speed of the plots is funt to me… Once I have this working, I was going to start getting more 20TB drives & increasing my farm.

I am creating regular plots on another system right now, just so I’m farming something while working on all of this.

Thank you for your help, I am starting to look at containers / liinked clones now.

I have the management GUI on its own physical NIC now, with nothing else on it. I now have the other three NICs added into Proxmox. The Plotter is on its own NIC, TrueNAS is too & so is the Farmer now also. There’s four ports on the NIC, so every main thing has its own port now. They are setup as Linux Bridge.

I’m working on setting up a container for the plotter, which I will create a template from once I have it initially setup the way I want. Thank you for your help! I’ll let you know how this goes…

What would be a point of that (having multiple NICs)?

The node has virtually no traffic (maybe 2Mbps). The plotter pushes data to storage only (on the same box). Harvester has way less traffic than the node (order of magnitude(s), I would think). Not sure how TrueNas is used, I mean only for the harvester data or also for your other things.

The point I am trying to make is that when you push right now data from your plotter to your TrueNAS, potentially all packets will need to hit physical part (go through the Ethernet cable). Although, I am not 100% certain about it (don’t know how it works in the context of VMs, i.e., which part of TCP/IP stack is sitting where).

On the other hand, if you just add all needed IPs to the same NIC, TCP/IP is smart enough to ignore your requests whether using IP or 127.0.0.1 and will shift the data on the driver level between those two VMs / containers.

One physical Ethernet port can have multiple IP addresses and when using IP bridging all is well controlled (for ages).

I was listening to jonesjr in this thread & I have the NIC card already in the server, so I used it. I’m new to Proxmox & Chia, so I’m just trying to get these compressed plots working in the best way. This is how I have the network setup currently:

Hey jonesjr, I created an LXC & got everything installed, but I don’t see how to attach the GPU to this container… I’m starting to look online how to get this done, please let me know if there’s an easy way or good resource you know of for this.

Thank you,

Unfortunately, I don’t know what that proxmox dialog is trying to convey. I mean, I understand those terms, but if my connotation would agree with what I see there, your original setup should work. So, some proxmox reading, maybe about hosting SSLed websites would help, as from any web server point of view the SSL problem will be exactly the same as you originally had.

I was just sharing the screen shot because I’m a very visual person & images help me understand faster sometimes. I’m going to go through the SonicWALL at my network perimeter with my IT company this week & see if there is something there blocking something. I’m also going to reach out to someone who used to work in my office who’s good with Linux & see if they have some time to look at what I have setup also. I’m working on getting the GPU pass through working in the LXC I setup from jonesjr’s advice.

Thank you everyone for your inputs

1 Like

I’m slowly chipping away at getting this to work. I have the plotter stable & creating C8 plots now. I needed to increase the RAM for the VM. I put using an LXC on pause until I have more time to figure out how to get GPU passthrough to work with LXC. I increased the RAM to 238418 GiB now (256GB converted into GiB = 238418). If anyone else finds this post with a similar problem, I got help on GitHub from madMAx43v3r directly a little bit & a few other people. Hopefully this helps someone else & saves them some time:

(PLOT #2) - Total plot creation time was 698.878 sec (11.648 min)

Host:
Dell R720XD
Dual E5-2697v2 12 Core / 24 Thread
512GB DDR3
GPU Passthrough - RTX 3060

VM:
20 vCPU
238418 GiB RAM
./cuda_plot_k32 -n 10 -C 8 -S 4 -t /mnt/pool1/ -d /mnt/pool1/ -c xxx -f xxx

You need to get the drivers on your proxmox host and then share gpu with container. Not sure if u need cudda library’s as I’m not familiar with this gpu plotting.

Allow me to guid.

Have a lot of good videos about proxmox and other server hardware
And but
To plot with mad max You don’t have to have a gpu tho. Just lots of cpu cores. And a place for the actual plot creation. (Ram or ssd)
15-20 minute plots with this method. I don’t know why anyone would want to plot faster tbh. Lol

I got everything working really well now. I’m up to a main powerful farmer, with two harvesters & around 10k plots, about to go up to 15k. I wanted to thank everyone in this thread one more time for everyone’s help. I’m about to post the question “Can I run Chia GUI and Gigahorse 2.4 on the same Proxmox VM?” I"m working on building a really good setup for the coming halving & future ones…