The Recoll indexing process, recollindex is usually configured to have very low priority and not disturb other activity on the machine. Still, on an idle system, even with multithreading disabled, it will use 100% of one core if needed and available. This may be enough to get a laptop fan to spin up in some cases. To prevent this, we want to limit the CPU utilisation for every short time quanta (e.g. not more than 20 mS for every 100 mS).
This would be extremely difficult to do from inside the indexing process, because of the many places where intensive CPU usage takes place, some not under our control (Xapian or external helpers). Happily enough, on Linux systems, you can use the cgroup facility to throttle the process CPU usage. This is further documented on the Recoll WEB site
