The default number of processes the worker starts equals to the number of cpus. Possibly as default this is to much. If there are many queues running in parallel and they are very heavy the machine hits its limits. Should we limit to the half by default?
I do not see why we should not try to take the most advantage of the machine by default but limit to half of it.
Half is stupid 
Need to check why all the threads in my scenario are going up to 100% but they are blocking all other stuff on the machine. I know that there is a setting in config to limit the processes. I was thinking about this issue because I discovered it by accident and how to prevent this case by default.
You can run trytond-worker with a smaller priority.
Ok - we can close this than. I know how to prevent this case, I only was thinking if there could be an issue with that. If not, fine.
May I know how to set the priority to worker to lower?