Check feeders and TM1 server recyle

[BariAbdul]

"You can only check feeders for a cell once per session. The action of
checking feeders actually feeds the components of the consolidation, so any
subsequent checking will not yield accurate results. If you want to check
feeders for a cell more than once, you must recycle the TM1 server before
every check"(This is described in Applix rules guide 9 as well as Developer guide 10.1.0)

Could Gurus please clarify what does it exactly means? Does it mean for subsequent feeders check afterwards ,Do I need to restart the server every time for correct results and why is it so? Really appreciate if someone could please explain it. Thanks in Advance.

[Alan Kirk]

"You can only check feeders for a cell once per session. The action of
checking feeders actually feeds the components of the consolidation, so any
subsequent checking will not yield accurate results. If you want to check
feeders for a cell more than once, you must recycle the TM1 server before
every check"(This is described in Applix rules guide 9 as well as Developer guide 10.1.0)

Could Gurus please clarify what does it exactly means? Does it mean for subsequent feeders check afterwards ,Do I need to restart the server every time for correct results and why is it so? Really appreciate if someone could please explain it. Thanks in Advance.

My understanding is that once a cell is flagged as "fed", that feeder flag doesn't go away if you subsequently reset the feeding value to zero; the feeder persists for the duration of the server session. (Remember that numeric feeders fire only once.) However I didn't want to come out and say that without trying to find something that backs it up. This is the closest I can find (my emphasis):

Once a cell is fed it will continue to refer to the original FEEDER even after the FEEDER statement is changed. There are two methods to force a reprocessing of the FEEDERS after a change has been made: 1) Issue a CubeProcessFeeders command in a TM1 TurboIntegrator (TI) process; or 2) Restart the IBM Cognos TM1 Server. This is critical to be aware of during the application development phase. For example, if an original FEEDER was over-feeding a cube but was then corrected by writing a more restrictive FEEDER that may appear to be accurate. However, until the TM1 Server is restarted it will not be possible to verify that the FEEDER is working correctly.

Accordingly if the act of doing the feeder check causes the cells to be fed, those feeder flags won't go away. Consequently if you recheck the feeder without a server bounce (or possibly reprocessing feeders, but I wouldn't count on that in this case) all of the cells are going to have been "fed" by the previous check, even though they may not really be fed by the rule feeders themselves.

[lotsaram]

"You can only check feeders for a cell once per session. The action of
checking feeders actually feeds the components of the consolidation, so any
subsequent checking will not yield accurate results. If you want to check
feeders for a cell more than once, you must recycle the TM1 server before
every check"(This is described in Applix rules guide 9 as well as Developer guide 10.1.0)

I think this is bogus and just another example of poor documentation. Maybe this was the behaviour in v1 of the rules guide with v6 of SinPer TM1 but I sincerely doubt it has been this way for a long time, if it ever was ...

I did a simple test of a view full of rule calculations which were not fed. Running check feeders on a cell resulted in a list of "not fed" cells as expected. Subsequent check feeders of the same cell resulted in the same list of not fed cells. Moreover turning zero suppression on in the view had the same result both before and after running check feeders, rule calculated cells did not show in the zero-suppressed view and non N level values above the not fed N rules did not start consolidating after the check feeders either.

Bogus.

[BariAbdul]

Thanks a lot, Alan and lotsaram.Much appreciated.