The way sling logging is setup you could either use scheduled rotation or size based rotation, both can not be used simultaneously in a single configuration.
There are more details around the loggers in sling documentation here
The best practice per say:
You need to have proper logging level set for the production instances so that the right amount of required logging happens, no noise is logged.
For instances lower than production it makes more sense to have size based rotation as those logs are usually required for debug purposes so the amount of logging tends to increase if you use daily rotation. And it becomes difficult to process those huge files with normal text editors.