THE CHIA FARM

[RimBlock]

Remote Chia GUI ??

Does anyone have a solution for being able to run the Chia GUI on a remote Windows machine and having the Full node running on a Linux machine ?.

At the moment, I have the Full node on a Windows machine inc GUI and I remote desktop to it every now and then to check and collect any farmed Chia.

This is problematic with Microsoft forcing monthly update reboots which have caused corruption with my SQLLite DB a couple of times now.

Has anyone set this up / come across a guide ?.

I only need access from my local lan and would rather not send info out to third party solutions.

Thanks

 

[Patrick]

So @Marsh - I was going to put a few 14-18 USB HDDs on a TinyMiniMicro node (I have a few extras) after using a PC with Optane for plotting. I told @WillTaillac and pretty sure he thought I was crazy for having such a small farm :-(

 

[mirrormax]

Its a fun hobby/learning project. just not very profitable, I set up a 500tb farm with a 24slot jbod and 10usb hdds back around launch, after plotting its low maintanance/low heat output compared to any sort of gpu farm of similar profitability. But still its currently less than a dollar a month per TiB. If i started now i think id just plot the Chives fork instead, similar profitability right now but not the same massive prefarm and more community driven instead of a corporation looking to IPO holding almost all the coins.

 

[Marsh]

Started around mid May 2021, about 1000 plots ( May to June ), slowly adding few hundred plots per month.
As of Feb 2022, total 8,600 plots ( 45 % Solo plots , 55 % Pool plots )

Total farmed 41 XCH, sold most of the XCH ( when price above $300 ) , keeping 16 XCH for now.

Just another way to help paying for cheap hard drives,
After finish farming, I would end up with ~1 PB media server.

 

[Rock]

Started in April 2021, 1,900 Plots, 4 XCH rewards.

 

[Marsh]

@RimBlock

I am using remote Windows GUI client connecting to a Linux Chia fullnode

Connecting the UI to a remote daemon

Connecting the UI to a remote daemon

Chia blockchain python implementation (full node, farmer, harvester, timelord, and wallet) - Chia-Network/chia-blockchain

github.com

 

[RimBlock]

Hmm, looks like the ChisBlockchain GUI needs to be installed on the Linux full node in order to generate the GUI ssl certs (used for remote GUI connectivity).

Having installed Ubuntu server, X-server, Gnome, VNC Server and got that mostly working, running the install-gui.sh throws up lots of the below errors (for different js files) ....

npm WARN tar ENOENT: no such file or directory, open '/home/chiauser/chia-blockchain/chia-blockchain-gui/node_modules/.staging/@material-ui/icons-3cf6e45d/esm/ZoomOutTwoTone.js'

There is no .staging under /home/chiauser/chia-blockchain/chia-blockchain-gui/node_modules/

It then falls over with an error trying to checkout master...

npm ERR! Command failed: git checkout master
npm ERR! error: pathspec 'master' did not match any file(s) known to git
npm ERR!

I am not a java dev so getting the dev env working is not something I am too familiar with. I can go through the install script but is it really this buggy trying to get the Chia Gui installed on Ubuntu ?

Any other alternatives for generating the GUI ssl keys without having to install the GUI itself ?.

Thanks

 

[Marsh]

Hmm, looks like the ChisBlockchain GUI needs to be installed on the Linux full node in order to generate the GUI ssl certs (used for remote GUI connectivity).

I did not install any Linux GUI on the full node.

Direction from Chia site
copy from linux host
~/.chia/mainnet/config/ssl/daemon/private_daemon.crt
~/.chia/mainnet/config/ssl/daemon/private_daemon.key

to Windows GUI host
~/.chia/mainnet/config/ssl/ui/
~/.chia/mainnet/config/ssl/ui/

make change to the Windows config.yaml
to reflect the Linux host IP

Linux host config.yaml
change to: from 127.0.0.1 to self_hostname: 0.0.0.0

restart chia full node

sudo netstat -tulpn | grep 55400
tcp 0 0 0.0.0.0:55400 0.0.0.0:* LISTEN

I just follow the Chia site instruction

 

[RimBlock]

My bad. Misread.

One clarification...

If I shut the GUI down on a machine running a full node it shuts down the full node as well.

If is shutdown the GUI on the Windows machine, does it still shutdown the full node on the remote Linux server ?.

Ok, to answer my own question... The remote gui does not shutdown the Farmer by closing it.

So I have it pretty much working. The Wallet needs to be re-synced as it didn't like the wallet DB from my Windows full node.

The harvesters all connected easily after copying over the ca certs.

I have swapped one of my plotter nodes from a X9DRT-HF to a X9DRT-HF+ (2x ram slots but no PCIe slots if using the SAS controller). Unfortunately the HF+ comes in a 1U tray whilst the HF came in a 2U.

I suspect the lack of a 2U baffle is causing the HF+ to lock up. I will have to fashion one myself. I do have a 2U HF+ but the sides of the tray do not have groves which are needed in order to use it in this chassis.

I could potentially run 4x plotter nodes with 2690v1s but considering the heat generated currently with only 2 nodes, I think it would be a significant issue in this tropical climate and that is without even thinking of the fan noise.

 

[Marsh]

If is shutdown the GUI on the Windows machine, does it still shutdown the full node on the remote Linux server ?.

No, my linux full node without GUI is always on

 

[RimBlock]

My ram upgrade came for my plotter today and as I am trying to fill 10x 10TB drives I thought I would have a bit of a play around.

I use a 2U Supermicro SuperServer 6027TR-DTRF 2-Node (e5 v1&2) with one x9-DRT-HF & one x9-DRT-HF+ node (I have another x9-DRT-HF+ spare). I got this as I have a fair amount for E5-2690v1s laying around and a bit of DDR3 ram that works, plus it was pretty cheap shipped.

Test plotter node 1 (x9-DRT-HF+)

• Ubuntu, 2x E5-2690v1, 256GB ram (1333MHz), 4x 400GB SAS SSD (Raid 0), Final plot disk - HGST 10TB SAS (HE).
• Note: this node has no PCIe slots but has 16 ram slots.

Baseline plot speed = 30min / plot
Usual Plot config = Tmp1 - 4x SAS SSD Raid 0, Tmp2 - 120GB Ram disk

Ran a couple of tests.

MadMax using only Ram

• Plot 1 - Tmp (only 1x 248GB Ramdisk)
• Plot time = 40 minutes / plot
• Assume the 1333MHz ram is slowing things down.

MadMax 2 plot stagger (17 min gap)

• Plot 1 - Tmp1 - Shared 4x SAS SSD Raid 0, Tmp2 - first 120GB Ram disk
• Plot 1 - Tmp1 - Shared 4x SAS SSD Raid 0, Tmp2 - second 120GB Ram disk
• Plot time = Baseline x2 = 60min. First plot = 39min , second plot = 47 min. First 2 plots 64 mins then another 2 each after 54mins (inc 7 min copy to completed drive.
• Occasional crashes of the second run. Possible CPU over-contention as all the drives have space (inc ram drives).

Might break up the 4x 400GB SAS ssd drives in to 2x 800GB Raid 0 arrays and see if that improves things.

Test plotter node 2 (x9-DRT-HF)

• Ubuntu, 2x E5-2690v1, 128GB ram (1333MHz), 3x 1TB NVME (no raid), Final plot disk - HGST 10TB SAS (HE).
• Note: 3x NVME slots but only 8 Ram slots.

Baseline plot speed = 30min / plot
Usual Plot config = Tmp1 - 1TB NVME, Tmp2 - 120GB Ram disk

Test run
MadMax using 2x NVME (raid 0)

• Plot 1 - Tmp1 - 2x 1TB NVME Raid 0, Tmp2 - 120GB Ram disk
• Plot time = 30 minutes / plot
• Seems 1x NVME can already keep up with the processes. 2x NVME in raid 0 adds no value for me.

Hope this helps someone.

 

[Marsh]

Plotter

Asrock X99 board ( approx $100 purchased from 2019 )
E5-4667 v3 , 16c / 32 T , all core turbo mod @ 2.9ghz ( $75 purchased from 2019 )
FusionIO iodrive2 1.2tb ( approx $100 purchased from 2019 )
128 gb RAM
Intel S3700 800gb SSD for staging of plot files , target drive for madmax plotter


23 min per plot , 60 plots per machine per day

Sometimes, I would run 8 plotters , usually 3 plotters at the same time.

 

[RimBlock]

Nice speeds. The good old days of cheep prices back in 2019.

Doing a quick search on ebay now that would be likely to set me back over $1.1k plus case & PSU etc although that does include shipping to where I am located which is not cheap...

I take it you are using pretty fast DDR4 ?. Trying to see why you would be faster with a slower CPU (suspect ram speed and destination drive) . What MadMax settings are you using ?.

Current cost of the 6027TR-DTRF 2-Node plus the E5-2690v1s, 128GB ram, 2x 1TB NVME etc comes in around $1,560 and gives two servers (48 plots x2 / day). These have no SAS drive compatibility, I had to get the SAS (midplane) cards separately.

I am considering going up to a 6028TR-DTRF which is around the same chassis price and runs on the Supermicro X10 (E5 v3&4) series but I then have to purchase the cpus etc from scratch again. Paying close to 2k for a few more plots per day is probably not the wizest move at this point. I do prefer the web based (rather than the X9 Java based) IPMI though.

I will try using the 2TB NVME I have in one of the machines as a 'done drive' and see if that helps things.

Update: no difference in plotting time but the 7 min pause whilst the plot is copied has now been reduced to 2 min.

 

[Marsh]

I take it you are using pretty fast DDR4 ?

Not fast ram, cheapest and slow 16gb DDR4 reg RAM that I could find in 2018-2019

 

[RimBlock]

Just for reference ...

Finished plotting my latest batch. Now at 277TiB (2800 plots). TTW flicking between 1m and 25d depending on the fluctuation network space.

To move forwards from here I would need to spend around 2k for more drives and a chassis (plus shipping etc). Not sure if it is worth doing over using the same case for some other crypto...

 

[RimBlock]

Mainnet space is going down. Down to 28.28 EiB today. 277TiB is now reporting a win every 23days.

New version of the Chia software (1.3.1) came out just over a week ago.

Chia has been bouncing between $60 & $70 over the las month.

Waiting for my next (overdue) win and then will possibly be moving some of my storage over to a file storage crypto or just selling up and purchasing crypto.

 

[EasyRhino]

Yeah I just realized that in my own rig my very expensive electricity made it a money loser to run my 4TB drives, and I unplug them.

Are there any other storage? Cryptos that return any sort of significant profit right now? Cryptos that return any sort of significant profit right now?

 

[RimBlock]

I have just started another thread asking the same question here .

 

[Bert]

How is Chia farming going on? Are you guys still in it?

My farm is still around ~254TB chugging along on 24 disks. Pretty disappointed that price is keep on sliding down; luckily netspace is going down with it, recuperating some of the loss income and there is no complete price crash. At so low prices, even for used SAS drives, it takes 3.3 years to take back the investment. I am not sure if I should keep hodling and or sell my chia to expand.

 

[msg7086]

Already converted my drives into a NAS.