K35 repeated plot creation failure

i fill my 18TB HDDs with 38 x K34 and 2 x K33 then the Drives has only 9GB - 10GB left, so u use 99.95% of the space.

1 Like

Yes, that is what I do.
And often, after having “38” K34 plots on one of my 18 TB drives, I end up 1 GB shy of fitting plot number “39”.

Sometimes I start to shuffle plots around, to fit K34 number “39”.

I copy the largest K34 files off of the drive to other drives that have a few GB of space, and take from those other drives their smallest K34 files (this involves a 3rd, intermediary, drive to hold the plots being transferred).

Each swapping of K34 plots results in saving 100 MB to, perhaps, 300 MB.
After several K34 plot swaps, number “39” fits on my 18 TB drive, leaving approximately 100 MB of free space.

I swore off trying to ‘swap & fit’ plot files after an initial frustrating attempt after making a bunch of K33, and some K34 plots. Why? I did a little calculation. I have 42 drives. If each one had 100GB of free space (hint: they don’t - the space varies from 100GB to a couple GBs), then I could reclaim 42x 100GB or 4.2tb of space, roughly enough for 40 odd k32 plots.

The hassle is that to move the current plots somewhere, plot better fitting k33, k34, k35, files, or move them from some other drive where they already live - is a royal PITA at 125-250MB/sec.

Then I could by a 4TB HD for what? Maybe $45-70 even new ! If I was really impoverished, that might dissuade me from buying… but I’m not. Or I could sell a 12GB and by a 16-18GB for about the same difference.

Although juggling plot files of various K’s is certainly an altruistic endeavor that I give heartfelt kudos to those that can suffer doing the work, it just doesn’t make a lot of sense as far as I can calculate.

1 Like

Agree, that looks like an OCD forced type work.

1 Like

No bad feeling behind that, as it is coming from my personal experience with OCD. :slight_smile:

By the way, if we assume that plot density is constant, sure you are adding extra hashes by closing those gaps.

On the other hand, those smaller plots most likely imply that they have less hashes, but most likely the same header/table structure. Therefore, those plots have a bit lower hash density. So, when you pack more of those smaller plots to fit all the available space, you may not be gaining any extra hashes.

So, this is another argument for “plot the HD” rather than “make one plot” plotter option.

Sounds like you might want to try PlotFS. It is beta software, but today it does work and uses FUSE to implement a filesystem consisting of many drives combined together into a single FS. If you lose a drive, you only lose the plots that were on that drive, and 1 or 2 plots that may have occupied the start and/or end of that drive.

Might be worth a look if you’re trying to minimize unused space.


I only have 40.9gb free on a 18tb drive with 165 plots
yea wish we could add up some spaces.

To put that in a perspective. Chia’s white paper advertises farming as kind of a hobby that can utilize overprovisioned HD space. That implies normal boxes that have 1-2 TB free space. With “make one fixed plot at a time” what is the overhead for such farm?

In the beginning I was keeping a spare plot of space on each drive, then folks talked about load those drives up to end… Will see what a 20TB disk will do when the price comes down to beer level…