2 nodes no more ingress

[Alexey]

Hashstore uses a probabilistic model to estimate a used space and calculate a free space in the allocation.
Similar to suggested there:

On a statistical approach for the used space file walker

We’ve recently got a large number of problems that require a full run of a used space file walker, and we have a large number of posts complaining about its duration. We have also seen that its duration is also a source of problems—not all changes to storage while the file walker is running are accounted. One key observation that was made is that the measured values of disk space do not have to be fully accurate, and it should be acceptable to have a decent approximation. This suggests that a s…

Seems the easy fix is to enable a dedicated disk, or use overprovisioning, the node would stop at 5GB of free space on the disk anyway. But it’s an emulation of a dedicated disk in this case, so why to make several hops?