Saving data with persistent feeders
The cube files are called cube-name.cub and cube-name.feeders.
When a cube is loaded, if a .feeders file exists, it is read to re-load the feeders and normal
feeder evaluation is skipped. If the feeders file does not exist normal feeder calculation occurs
and the .feeders file is written after all the feeders have been calculated. This is the behavior
the first time the
TM1® server
is brought up on a data
directory with the PersistentFeeders=T. As part of the server startup the .feeders
files for all the relevant cubes will be written. A new persistent feeders files is written for any
cube for which the persistent feeders file is not found at server startup. This means that the
.feeders files may be deleted from the data directory to force a complete re-calculation of feeders
when the server next starts. For more information on how to delete a feeder, see Deleting persistent feeders.
The modified time of the .feeders file is compared to that of the base cube file at load time. If the cube file is newer than the feeders file, the .feeders file is regarded as invalid and is deleted. The .feeders file is always written after the cube file so it should always be newer than the cube file.