Refile is an event-driven service. Whenever Refile detects an event trigger, such as a change in the default security or metadata for a container or a document, a refile event is queued to be run. These events form a first in, first out event queue. Each refile event must be concluded before the next queued event begins. Refile automatically continues processing each queued event.

Changing the security status of a container, a document's default security, or an eligible metadata refile field triggers a refile event and is the starting point of the refile process. The Refile Service processes items in a downward flow, through each subfolder and is recursive through all the subfolders. For example, a change at the workspace level may affect all items within that workspace. If the event occurs in a child folder within that workspace, only that child folder and items contained by it, including other folders, may be affected. The propagation never traverses to a higher container. The operation occurs in the background and is not noticeable by users. They can continue using the application. However, it may take a few minutes for the refile process to propagate these changes through the entire container hierarchy.

Refile is a stateless operation. That is, when Refile is attempting to update information (either for security or metadata) the original information, or the state of the item is not recorded. Only the information that is available at the moment for the refile operation is considered, and not the information that was changed.

Refile includes robust error handling. If a refile event does not complete successfully, such as due to a network interruption, it will automatically resume and continues from the point where the interruption occurred.