Problem is than Data Ingres is not shared logicaly.
Some nodes has 25-30GB ingress, other has 500mb ingress. Thay sharing 1 connection.
some difference, then slow node was full for some time, and now as olld data deletion has new free space. some has 100GB free other has 1,2TB free after deletion.
So there is big question is there some bug that not allow new data income?
There have been a few code changes in node selection, but as far as I can tell nothing that would select one node over another, unless one of the nodes is new.
As this nodes was previos full, they not not got data from new test sattelite.
So thay are not veeted yet on new test sattelite. But other that are newer and had space already veeted.