Trace Feeder and Check Feeder Disagree

[asegal]

I have a very simple rule and feeder like such:

['Total Departmental Income (Loss) - 000'] = ['Total Revenue - 000'] - ['Total Departmental Expenses - 000'];
['Total Hotel GOP - 000'] = ['Total Departmental Income (Loss) - 000'] - ['Total Undistributed Expenses - 000'];

Feeders;
['Total Revenue - 000'] => ['Total Departmental Income (Loss) - 000'];
['Total Departmental Expenses - 000'] => ['Total Departmental Income (Loss) - 000'];
['Total Departmental Income (Loss) - 000'] => ['Total Hotel GOP - 000'];

This is all within an account dimension. I'm having an issue where Total Departmental Income (Loss) is not being fed, and thus Total Hotel GOP is not being fed when Total Revenue - 000 is zero but Total Departmental Expenses - 000 is not zero.

Trying to trace why this is not working, I check feeders on Total Departmental Income (Loss) - 000 and it says that the cell is not fed. But when I trace feeders on a singular leaf cell part of the consolidation Total Departmental Expenses - 000 it says that Total Departmental Income (Loss) IS FED as is Total Hotel GOP. Why would check and trace feeders disagree about the exact same cell intersection being fed?

2024-08-30 11_05_32- Cube Viewer_ hAVE-_hAVEFinCube-_Default.png

2024-08-30 11_06_25-Rules Tracer_ hAVE-_hAVEFinCube.png (55.78 KiB) Viewed 3386 times

[asegal]

I will add a little bit more context here after further discovery.

1. Doing a CubeProcessFeeders works and the cells are fed and rules calculated.
2. The values are loaded via TI.
3. The element that is fed from (LH side of feeder) is a "new" leaf element that is inserted by TI process when loading data. Basically the TI when loading checks whether the leaf account exists in the dimension and if it does not, inserts it, and then loads the data to the new element. Later on, we "move" the new element from an unclassified holding consolidation to its correct place (in the example above, as a child of Total Departmental Expenses - 000).

So my hunch is that if you move an element that already has a value into a consolidation that is used on the LHS of a feeder, it doesn't fire correctly because the cell value is not changing after it's been moved into that consolidation.

Does this seem right?

[asegal]

I think I answered my own question here. Seems in line with viewtopic.php?t=311

I speculate that "Trace Feeders" shows what "should be" fed based on the setup vs. "check feeders" shows what IS fed based on the last time the feeders fired.

[Steve Rowe]

You've got to the right answer.

It's important to remember what triggers a feeder to execute as they aren't live like a rule is.
When a server starts (assuming no persistent feeders)
When data changes from null to non-zero
Forced recompile via TI.

Changing the structure of a dimension doesn't force a recompile of feeders (thankfully). This does mean that if you have C levels in the left / right hand side of the feeder statement then you need put new members in there before you add data.

Cheers

[asegal]

Thanks. And am I correct on the behavior of Trace Feeder vs Check and why they would disagree?

[Steve Rowe]

Yes, also correct, Trace shows what would happen now if the feeder statement fired and Check shows what did happen.