0.3.1 over commits DST

[syadnom]

I have a number of dst drives, when they get down to <200GB free, plotman will still target them for new plots multiple times. Ie, 138GB free, plotman will setup 2 inbound plots to the drive

Plotman should count the number of inbound plots and reduce that from the GBfree value before selecting that dst for a new plot.

[angryBOT]

That is not a bug, but expected behaviour. It is the first question answered in the FAQ.

[syadnom]

I misunderstood. thanks

[syadnom]

I'm revisiting this because I think it is a bug. Maybe the code is doing what's expected, but that expectation is a 'bug'. filling up dst disks ends plotting, hangs the whole process up. If archiving bothers to check free space then plotting dst should as well. IMO.

[altendky]

dst is meant as a buffer. It is intended to be continuously emptied by either archiving or your own alternative. As such, the available space on the dst drives when a plot launches is expected to have little to do with how much free space there will be hours later when the plot is finishing and needs space on the dst drive. Archiving is the intended means for distributing plots to fill up drives.

That said, setting up local archiving has been needlessly painful since it depends on having setup rsyncd regardless. There has also been lots of understandable confusion that archival is only relevant when the drives are not local. I have reworked the archiving setup because of this and on the development branch you can now use archiving for local drives without rsyncd. This is expected to be released in the next few days, hopefully.