TM1 Forum

Hi Expertos!

I just want clarifications.

We had this data validation. My user gave me two rows from his own-created Sales Summary report. While my TM1 Sales Summary cube returned only one row. We are sourcing from same database table. So I checked if there's wrong in my Rules or if something was just not fed.

We are using Tm1 10.1.1

Until I opened the Rules editor of Sales Detail cube (Sales Summary gets data from Sales Detail via Rules). Then I inspected if I missed something. Nothing was wrong. What I did is to hit Enter button. And Saved. Then my Sales Summary cube showed two rows.

First thing I thought was, the Sales Detail.feeders file was outdated when compared to the Sales Detail.cube file.

1. But how can a feeder file be outdated? (although I didn't really see the datetime of the two objects, just my suspicion)

2. And does the feeder file store ONLY the feeder value? Or the feeder variable?


For example, I uploaded (vai TI) sales qty for the month of JAN.

Month Product Qty and Cost are from Database table. Sales is in TM1, rule-based. (So don't get confused with the SALES column in the sample)

MONTH | PRODUCT | QTY | COST | SALES

JAN | MANGO | 0 | $5 | 0
JAN | APPLE | 4 | $5 | $20


Came FEB, I uploaded sales qty for FEB. Without shutting down the TM1 server instance from the time uploaded JAN data up to the time I uploaded FEB data.

MONTH | PRODUCT | QTY | COST | SALES

FEB | MANGO | 4 | $5 | $20
FEB | APPLE | 3 | $5 | $15


3. Does the Feeder file know that it has to feed the SALES for the Month of FEB? (because now the Qty is not zero anymore)

4. Do feeder files need to be deleted every server startup?


Thanks in advance.

bunchukokoy wrote: 1. But how can a feeder file be outdated? (although I didn't really see the datetime of the two objects, just my suspicion)

2. And does the feeder file store ONLY the feeder value? Or the feeder variable?

3. Does the Feeder file know that it has to feed the SALES for the Month of FEB? (because now the Qty is not zero anymore)

4. Do feeder files need to be deleted every server startup?

1/ in 9.5.1 and 9.5.2 feeder files could be outdated if rules did not use skipcheck & feeders or if feeders were populated from a feeder statement in another cube. As per my understanding this should be fixed in 10.1 and 10.2 and the feeder file is updated on disk whenever the cub file is saved so a feeder file can't be out of date in this sense.
2/ the feeder in nothing more that a marker to indicate whether the cell should be consolidated. You can think of it as a 1/0 boolean flag
3/ We're not clairvoyant. You would need to provide the feeder statement so someone could judge if the feeder is written correctly
4/ No. The reason it is recommended to periodically delete feeder files is to prevent the accumulation of feeder pointers which are no longer relevant (that is cells with a value of zero which are marked by feeders to consolidate) the reason for this is to prevent performance degradation of rule consolidation due to overfeeding.

Persistent feeders do not change the normal action of feeders "firing" upon new data entry and causing new areas to be fed. By the sounds of it your issue is that somewhere a "condition flag" in either an attribute or control cube has changed that is used in the feeder evaluation to "redirect" a feeder, thus that the area on the left hand side of the feeder statement has already done its feeding but were it to be re-run it would feed somewhere else. Remember numeric feeders only fire once when the cells change from 0 to anything else. In this instance you don't need to delete all feeders and restart the server you should just run the TI function to reprocess feeders (manually saving the rule file has the same effect).

Thanks for explaining it well Lotsa!