Hello, my 4TB node is getting full, I’ll add another 6TB disk I have (I’ll crate new node) When I have 2 nodes on one /24, transfer is 50%:50% divided. But what happens when one node is full? Are satellites smart enough to send 100% ingress to ‘not full’ node?
In other words: is it better to migrate 4TB=>6TB disk and add new 4TB node, or no need to bother?
Another question: what % of disk capacity free consider as disk is full, it's time to add capacity?
Edit: GB → TB fixed
The very short answer is yes 
The longer answer is, that there are many parameters defining how much traffic your node (either one) will receive. Latancy, disk responsetime, cache settings, % online, % audit etc. etc. All used for reputation calculation of each node and the corresponding IP it’s operated behind.
But in general, if they perform the same, you should see ALL traffic go to the one with free space.
I think it’s very individual, I have not added nodes to single IP’s for a long time. For the past year or so, it’s been a rollercoaster ride of up’s and down on capacity usage. Currently most of my nodes are at 65-80% used, and I’m not yet considering to extend them (ZFS pools).
Just start another node on the same IP, on the new drive.
You should aim for a date when your node fills, the new one to be fully vetted. Vetting process takes a month or two, so think how much space will fill in 2 months. That is your limit for left free space. Maybe aim for 3 months just to be sure.
… and you should correct the GB in the first post. You ment TB. 
Vetting is done in 1 day for US1 atm, 3…5 days for other satellites.
Stupid me 
Thanks.
And thanks all for answers.