Well in working with scrum it's always seen as important to make clearly defined stop and start points of small tasks.
However how do you capture "ongoing" tasks into a scrum system?
Like on friday after deployment we typically have "work on documentation and code review of old code to check for bad practices". This has no end, as constantly new code is found and new practices emerge; require updating of documentation and blaming old code to be updated. We're also too small of a company to make a separate system for these (devops and the likes are still tightly coupled to development). So to provide the manager of a time frame these things are calculated into the scrum work.
However how do you suggest to "mark" them - so that both management knows what happens, and we do not have constant backlogs of the same wording.