When I'm writing to the user share instead of directly to a target disk using "most free" it is writing to disk1 which is basically full instead of to disk 9 which is empty.  I don't have any specific drives allocated to the user shares, expecting them to span them all.  This sounds exactly like the "high water disk full" thread as well.

 

Probably what's wrong is that 'split level' for the share is not high enough - if set to the default value of 0, then the share will not span multiple disks at all.

 

Which brings me to this: In the next release we are going to change the meaning slightly of 'split level'.  If you leave the split level value at 0, then this will function as if you set it to a very high value now, e.g., 999.  That is, any time any object is created at any level of the share directory hierarchy, the disk it will be created on will be selected according the current 'allocation method'.  This way, shares will fill up across disks without additional user intervention, and the only time you have to worry about split level is for special circumstances such as wanting to guarantee keeping all files of media directory on the same disk.

 

This is way more intuitive and we should have done this to being with (slaps forehead).

Did the split level logic change with 4.3?  Only reason I ask is that I was using the defaults in 4.2 and it seemed to be spanning fine (and yes, they are all set to the default of 0).