Chia (XCH) Discussion Thread

[Kilrah]

I keep moving plots around, connecting and disconnecting drives, reaching some plots through the network etc, no problem and I haven't typed a single command line? Must have done something wrong with your wallet.

 

 

[Hill160881]

7 minutes ago, Kilrah said:

I keep moving plots around, connecting and disconnecting drives, reaching some plots through the network etc, no problem and I haven't typed a single command line? Must have done something wrong with your wallet.

 

 

The power went out. I made no changes.  10tb plotted and now I can’t sync any of it.
I am going to try upgrading to 1.1.2 and see what happens. 
 

Also I am now thinking it’s my network. It’s huge with 3 routers. I also see you can’t run two machines on the same network without some porting workaround so that may have also caused an issue. It just so new and the GUI is not great.  Then there is the fact I am better with tools than keyboards lol. In reality I don’t know crap about networks, how to set them up, how that will interact with chia.....  going to be a long road for me haha

[Hill160881]

22 minutes ago, RollinLower said:

i mean, posting two random posts with just complaints isn't helping the situation.

i've had my system rebooted, power lost and just unplugged from the wall a couple times, and the only plots i lost where the ones that where still plottng. all my farming plots are intact.

seems like you just got unlucky somehow here man.

Just really frustrated. Many are having this issue and no one has a handle on why this is happening or how to fix it. Syncing issues are a problem.  It seems to be a porting issue but why a power outage would mess it up is a mystery. It may just be quicker for me to start over and put the full system power into plotting to catch up instead of it running them in the background. 
I also can not get my second machine to connect on the same network as the other one and still can’t understand the instructions on how to get this fixed. 
 

If I get it figured out and understand how I fixed it I will put together a thread on it. But usually I fix stuff like this and have no idea how I did it lol. 

[BondiBlue]

1 hour ago, Hill160881 said:

The power went out. I made no changes.  10tb plotted and now I can’t sync any of it.
I am going to try upgrading to 1.1.2 and see what happens. 
 

Also I am now thinking it’s my network. It’s huge with 3 routers. I also see you can’t run two machines on the same network without some porting workaround so that may have also caused an issue. It just so new and the GUI is not great.  Then there is the fact I am better with tools than keyboards lol. In reality I don’t know crap about networks, how to set them up, how that will interact with chia.....  going to be a long road for me haha

So you're having issues syncing? That's more of a network issue than a plotting instability issue. There are plenty of resources that tell you how to manually sync it. Once you do that it should be perfectly fine. 

[Kilrah]

It's not clear whether his issue is syncing to the network (as the name would suggest) or the plots not being recognised anymore by the app for farming (what seemed to be from the explanation)...

[Hill160881]

So somehow my TCP was closed when the power was tripped. I had to go and open the tcp in network settings like when I set it up initially. No clue why this happened.
 

Completely out of my depth but it was a network setting in windows that changed.  
 

Why cant these things be fixed with a wrench instead? Much easier. 

[Kilrah]

Grr lost 3 plots at 40%, PC froze when opening HwInfo to check SSD temps

[BondiBlue]

4 minutes ago, Kilrah said:

Grr lost 3 plots at 40%, PC froze when opening HwInfo to check SSD temps

So far I haven't lost any, but I have a few that are stuck in the void somewhere. I accidentally started two plots on the same 256GB SSD and it filled up. I had to erase the drive to free it up, and since then those two plots have still been making slow progress somewhere. 

[RollinLower]

21 minutes ago, Hill160881 said:

So somehow my TCP was closed when the power was tripped. I had to go and open the tcp in network settings like when I set it up initially. No clue why this happened.
 

Completely out of my depth but it was a network setting in windows that changed.  
 

Why cant these things be fixed with a wrench instead? Much easier. 

i mean, if it's getting to your nerves there's always the percussive option:

[Hill160881]

Had to disable upnp to get two machines going on the same network. Does this make sense to anyone here?  I let my brother take over at that point to avoid the hammer solution. 

4 minutes ago, RollinLower said:

i mean, if it's getting to your nerves there's always the percussive option:

Was considering this after three days of no success. I still have hair so just in time. 

[RollinLower]

21 minutes ago, Hill160881 said:

Had to disable upnp to get two machines going on the same network. Does this make sense to anyone here?  

Yeah i've been there. When i started farming Chia it was on a single server in my setup, then i added a big storage server, and when that got too small i swapped the whole setup out for a dedicated plotting machine and a JBOD with 14TB disks.

every time i swapped machines getting the plots over has been a fight.

[2FA]

This is where I am currently, getting two more 18TB drives so that should be a couple weeks of plotting I think. Netspace has been going down since yesterday's network issue too so better chances at a block than a few days ago.

 

[RollinLower]

46 minutes ago, 2FA said:

This is where I am currently, getting two more 18TB drives so that should be a couple weeks of plotting I think. Netspace has been going down since yesterday's network issue too so better chances at a block than a few days ago.

 

ooh, that's pretty sick!

i'm still sitting on 2 XCH mined for now, but i only just broke 100 plots aswell. i got 6 more 14TB drives that need filling up so i definitely have some plotting to do.

 

on that note, what is considered best parctice? i got a 14 core CPU plotting in the dedicated machine. should i just do 14 parrallel plots until they finish, and then 14 again. or should i break it up in smaller batches (3 plots on one disk, 3 plots on another, etc)

[Kilrah]

I'm at 110 plots, about 25/day now as expected. Tallied up everything and I'll have about 70TB total, and it should take me another 3 weeks to fill these up. Network size drop was a welcome sight since it's been climbing so fast. 

 

 

 

27 minutes ago, RollinLower said:

on that note, what is considered best parctice? i got a 14 core CPU plotting in the dedicated machine. should i just do 14 parrallel plots until they finish, and then 14 again. or should i break it up in smaller batches (3 plots on one disk, 3 plots on another, etc)

Since the SSDs are being hammered I go for one simultaneous plot per SSD, with the exception of one of my builds that has 2 RAIDed NVMes since they essentially get half the access each so I do 2 of them there.

 

Then split the threads + 1 or 2 based on the number of simultaneous plots and give a bunch of RAM. I'm set at 3 simultaneous plots, 6GB RAM and 6 threads per plot on each 8-core machine.

 

I've just had to bring down the 5960X for maintenance but fortunately its 3 plots finished close to each other, so only lost an hour or so... ETH mining has surged so much on NH that it was worth installing my 1070 and start running it too, usually isn't worth it.

[RollinLower]

11 minutes ago, Kilrah said:

I'm at 110 plots, about 25/day as expected. Tallied up everything and I'll have about 70TB total, and it should take me another 3 weeks to fill these up. Network size drop was a welcome sight since it's been climbing so fast. 

 

 

Since the SSDs are being hammered I go for one simultaneous plot per SSD, with the exception of one of my builds that has 2 RAIDed NVMes since they essentially get half the access each so I do 2 of them there.

 

Then split the threads + 1 or 2 based on the number of simultaneous plots and give a bunch of RAM. I'm set at 3 simultaneous plots, 6GB RAM and 6 threads per plot on each 8-core machine.

so generally speaking, if i split up my 10 SSDs in this machine into 5 RAID 0's, i would see an improvement in plotting time, right?

[2FA]

24 minutes ago, RollinLower said:

ooh, that's pretty sick!

i'm still sitting on 2 XCH mined for now, but i only just broke 100 plots aswell. i got 6 more 14TB drives that need filling up so i definitely have some plotting to do.

 

on that note, what is considered best parctice? i got a 14 core CPU plotting in the dedicated machine. should i just do 14 parrallel plots until they finish, and then 14 again. or should i break it up in smaller batches (3 plots on one disk, 3 plots on another, etc)

Use this to automate it https://github.com/swar/Swar-Chia-Plot-Manager. It's written by one of the mods of the Flexpool discord. Also generally if you're doing multiple plots per drive, you don't want to exceed ~80% of the capacity or else it really slows down plotting and I also stagger plots on the same drive by an hour. Also see this graphic for the resources used during each phase

 

I generally set the temp2 directory the same as the destination directory since my destination drive is local and I move it to my actual harvester once full. This turns the final plot copy into a quick rename instead of spending 15 minutes or copying.

 

 

[Hill160881]

I am currently plotting on my 3900x system which is a little slow. I am also currently plotting on 8 stripped 4tb drives instead of a ssd which is not the fastest but not slow. Also I am plotting on my eth mining rig as well because why not. They are not so powerful but 10 plots every 24 hours is more than I was getting so...
However I am building a new system this weekend to help plot out the storage I now have. I am at 20tb plotted but I have 70tb currently with another 160 on the way. However I need more plotting power so I am getting a threadripper. I plan to run the 24 core processor with four 2tb m.2 drives and 128gb 3600mhz ram. Should plot fast enough but still going to take a while to plot out over 200tb. I also stagger my plots and assign about 6gb memory per plot. I should be able to run 20plots at once while mining eth on  4 gpus. For the record I orders all of this before anyone had a video on it. 
I considered a server system but no one in my family has any experience and everyone wants to keep it on home computers and a large nas for now. Should be a fun weekend. 

[tikker]

Dang if only I could empty my NAS and dedicate it to this.

[Hill160881]

I am wondering what the difference in drive life span of plotting on a stripped hdd setup vs m.2 SSDs? What if I stripped up ten 1tb drives for speed and plotted on that vs four m.2 drives. Will the hdd raid fail as fast as the m.2 drives? 

[BondiBlue]

26 minutes ago, Hill160881 said:

I am wondering what the difference in drive life span of plotting on a stripped hdd setup vs m.2 SSDs? What if I stripped up ten 1tb drives for speed and plotted on that vs four m.2 drives. Will the hdd raid fail as fast as the m.2 drives? 

It's definitely putting some work on those hard drives, but they should be fine. Plotting on mechanical drives (even in a striped array) will always be slower than an SSD though, so once you've got an SSD to use you'll be good. 

[Kilrah]

Well it's going to have worse performance since what matters is random access more than speed, and... well with an SSD you can know where it is at in its life unlike an HDD. 

[Hill160881]

I was just hoping to find a long term plotting solution but it sounds like ssd is really the only way. 

[Kilrah]

Well, a few 512GB SSDs every few months are peanuts next to a TR server and dozens of TB of storage, lol.

 

 

[BondiBlue]

1 hour ago, Kilrah said:

Well, a few 512GB SSDs every few months are peanuts next to a TR server and dozens of TB of storage, lol.

Exactly. Plus, SSDs have been so cheap for so long that it's now possible to get them for free from old computers. One of my plotting drives is a 256GB Toshiba M.2 SATA drive that I pulled from a scrapped Lenovo AIO. I don't know how much life it has left, but it's fine for Chia. 

[Hill160881]

So my eight 4tb wd red stripped raid setup running 1000Mb/s 32tb total, is plotting 30plots in 24 hours plotting on the raid HDDs. 3.3gb ram per plot. No ssd. Leaving 2 cores available for eth mining in the background. 
 

How does sound to others?