How does one point to multiple ( ideally) external drives ( and telling the software to max out the plot size)?
Another ‘similar’ question - of the same category at least:
Suppose you changed your base path to an external. You then plotted 10T, but the drive is 20T in size.
Can you ‘restart’ the farmer, asking it to use 15T (i.e. larger than initially requested) or would once need to do a wipe all the plots?
Currently there are two ways to go with multiple drives. Either run a node/farmer instance for each, or merge all drives in RAID0 or a filesystem that supports multiple drives like btrfs.
What’s the best approach is debatable. If you run multiple nodes it can be resource intensive - for the stress test I had 16 nodes running on a Core i7 10th gen machine with 32 GB - RAM was at the limit, CPU was fine.
Plotting seems to be quite resource intensive though, i see all 8 cores reaching 100% when plotting, so my guess is you would have to sync each node/plot in sequence.
If you go with merging drives, that will be a leaner setup in the end but my understanding is that sync/plotting time grows with the plot size, and possibly you hit a limit where sync can’t keep up with block production - I’ve tried plotting 10TB on an old Core i5 and that didn’t work at all. Anyway I would love for the team to chime in on this.
About your second question, if you wan’t to change the plot size you do have to wipe and replot.
Thanks so much for this.
I’ve got a couple of systems running at the moment - experiments really. One being a couple of drives connected together via LMV. Together I’m got ~16T (2x8T) ironically on an i5 - and I saw my CPUs peg out.
Then on another system, just a single 14T drive, going to see what happens between these two setups. Will report back anything of interest. I will say this, it is taking a long time to sync - I’m on #1100 (or so) after a couple of hours. Don’t know what I’m doing wrong, but can see others in the 100k’s via the explorer.
Great answer, yes you covered everything in regards to multiple drives and their options.
To your question about plotting and sync limit, these are both things that are going to be improved as we continue to progress. For example plotting will be significantly lighter weight once we resolve a few issues.
We are also planning to minimize the node dependency which will also free up system resources.
Theoretically we want to have no limit on plot size (other than what your System/OS can handle)