Whow knows how to plot to multiple drives?

[Cminer3207]

Somebody out there has figured this out...
How do you get the thing to fill up a drive and then move on to a new drive without stopping, changing the YAML, and starting again.

[jhony633]

I also seen this on another user but can't figure it out

[Cminer3207]

Here is the answer:

In config.YAML File...

At bottom under job...

Single Drive format:
———————————
destination_directory: J:\Plots

Multiple Drive Format:
——————————
destination_directory:

• J:\Plots
• K:\Plots
• L:\ Plots

Happy mining.

[paullepa]

For posterity syntax is:
(space)(space)(dash)(space)(DriveLetter)
ie.

• J:\

or

• J:\my_temp_example

I'm new to YAML and it took me forever to figure out that YAML parser sucks and needs very well formated text.

[Cminer3207]

You’re a good person paullepa...

That (space)(space)(dash)(space) took me three attempts to figure out...

If Swar had taken the time to say that he could have saved me some headaches...

I can’t complain too much though... he did make the plotting tool for us all...

[Kali123411]

For posterity syntax is:
(space)(space)(dash)(space)(DriveLetter)
ie.

• J:\

or

• J:\my_temp_example

I'm new to YAML and it took me forever to figure out that YAML parser sucks and needs very well formated text.

hmmm this is not working for me..

[paullepa]

try using yaml validator - it will point out for you where its having issues.
http://www.yamllint.com/

I'm new to YAML as find it very very picky

[Cminer3207]

Use the default YAML.
Second job example.
Copy and paste the Destination Drive exactly as shown there, then change the letters to yours.
If you have more drives, just paste it more times.

[dssamusaran]

With this multiple drives configuration, is the system handling full drives properly?

What I mean is: If a drive is full it then chooses the next one automatically.

Or do you still have to remove the full drives from the configuration?

[Cminer3207]

Thank you for sharing this.

So

Each Destination Drive (or group of drives with the same available space) needs its own job-order.
Set the max_plots: parameter to the size of the available drive-space.
maxPlots=(AvailableDriveSpace/108GB)

This will stop creating plots just before the drive space is full and not cause a log-jam in your max_concurrent or require deleting temp storage.

[dssamusaran]

Yes this is exactly what I did. A job for each size of drives and having the max job stopping before filling up. I left around 3 plots of margin for each drives to be safe. I'll fill them up by moving files manually at the end.

[Cminer3207]

Please post a description of how you move plots between drives.
I have heard is not as simple as just copying the .plot file to a new drive and erasing the old one.
Thanks.

[dssamusaran]

At first I simply cut/paste from source to destination. But it happened sometimes that the moved plots were no more recognized until full restart of the chia node.

To avoid this I created folders on each drives called 'ChiaPlotInput' and 'ChiaPlotTransfered'.
Those drives are NOT checked by chia.

Here's an example of moving plots from D:\ChiaPlot to E:\ChiaPlot

-Copy the plots you want to move from D:`\ChiaPlot to E:\ChiaPlotInput (this is the long part to copy the datas)
-Once done, Move the same plots from D:\ChiaPlot to D:\ChiaPlotTransfered (same drive, instant)
-In the Chia software in the Plot tab refresh the plot directories. The plot you copied will disappear from the farm. (could be done in commande as well following what you use)
-Move the plots from E:\ChiaPlotInput to E:\ChiaPlot (same drive, instant)
-In Chia, refresh the plots and they will come back, now on the correct drive.
-Delete the plots from D:\ChiaPlotTransfered

The whole 'ChiaPlotTransfered folder is just to be safe. You could skip that and delete the plots immediately but I prefer having a failsafe in case I make a mistake. I only delete the plots from there when I see all my plots being farmed.

[Cminer3207]

Brilliant!

[Cminer3207]

It fills all the drives concurrently, instead of filling one completely and then moving on to the next.

D:\ 2 plot
E:\ 1 plot
F:\ 1 plot

etc...

[Kali123411]

It fills all the drives concurrently, instead of filling one completely and then moving on to the next.

D:\ 2 plot
E:\ 1 plot
F:\ 1 plot

etc...

Yea that's why I just made separate jobs for each one of my drives.

[Cminer3207]

I prefer it this way.
I use jobs to split the temp space and cpu power into scalable worker units based on the most efficient configuration for system hardware.
I give them a weeks worth of drive space (252 plots) connect a UPS, and forget about it as I move on to the next build.

Changing the drive space is the only thing that bolloxes up the manager on a restart... so I give it plenty and tweak the other config parameters until it’s optimal.

[mkirshner]

Currently my setup is doing what the OP says- Swar assigns all parallel plots to one drive, fills it, then moves on.

Any advice on how to achieve the opposite- one parallel plot goes to one destination drive? I want to reduce the bottleneck from writing to the hard drive. Thanks