Version of Planning Analytics? I would retest the same scenario in 2.0.9 on a test environment. Depending on your version it's possible this is a defect that has already been resolved.
There should be no difference between the following three things:
1 - Running the CubeProcessFeeder Ti function
2 - Updating the feeder statements in your cube rules and saving
3 - Starting the TM1 model without persistent feeders enabled or removing the persistent feeders (.feeders files in the data directory) and starting the TM1 model when persistent feeders are enabled
Having PersistentFeeders enabled should have no impact on the
CubeProcessFeeders Ti function. This parameter should only impact the TM1 model startup.
Also, out of curiosity, what happens if you run a CubeUnload Ti function followed then CubeProcessFeeders. I believe the CubeUnload Ti function should remove all the feeders on the cube. This might more equivalent to modifying the rules (e.g a workaround in TI).
------------------------------
Stuart King
IBM Planning Analytics Offering Manager
------------------------------
Original Message:
Sent: Wed February 19, 2020 05:20 AM
From: Amarins van de Voorde
Subject: CubeProcessFeeders vs feeder files: feeder not fully updated after metadata changes
Hi Ryan,
thanks for your quick response!
1) By metadata changes I mean new dimension elements, mostly.
2) The consolidation of a new element doesn't work. But when I check the feeder (in the "source" cube), it does point to the correct cell. And the cells in the "target" cube don't say they're not fed. So the changed consolidations should work, but they just don't.
3) I expect the feeder files not to be complete/fully updated after running CubeProcessFeeders, as the changed consolidations don't work properly. SaveData doesn't fix them, we've tried that. The only thing that seems to help is to force the feeder to recalculate, i.e. removing the feeder and putting it back again.
4) We haven't been able to test whether turning off persistent feeders helps, as it's a production model and it's used a lot. That would be my next step, though. Maybe we can recreate the issue on development and try turning off persistent feeders there, I'll look into that.
------------------------------
Amarins van de Voorde
Original Message:
Sent: Wed February 19, 2020 03:51 AM
From: Ryan Clapp
Subject: CubeProcessFeeders vs feeder files: feeder not fully updated after metadata changes
Hey Amarins
A few questions for you.
1) when you say metadata changes, what is changing exactly? New elements, new attribute values, etc?
2) when CubeProcessFeeders doesn't work as anticipated what goes wrong? Are values left unfed and consolidations don't work?
3) how do you know the feeder files on disk are not right? Does a SaveData fix them?
4) if you turn persistent feeders off does it behave correctly?
Thanks!
Original Message------
We've noticed that, since we started using feeder files, CubeProcessFeeders doesn't seem to fully recalculate the feeder when there's a metadata change. CubeProcessFeeders was always the safe way of 'forcing' the feeder (up until TM1 10.2.2), but it doesn't seem to do the trick anymore. Only steps like removing the feeder line(s) seems to help.
We're not using conditional feeders, by the way.
We're trying to find a 'safe' solution, so our key-users can force the feeder again after a metadata change. Having them clear out the rule and then re-loading the rule file isn't 'safe' (for them), in my opinion.
Has anyone come across this problem as well and did you find a proper solution?
------------------------------
Amarins van de Voorde
------------------------------
#PlanningAnalyticswithWatson