That's interesting. Need to keep gathering facts 
I would do a sudo systemctl status nginx.service in between each command to watch for messages and keep track of the pids that are active (in the CGroup). Or, a briefer display of pids using something like: ps -eF | grep -E "nginx|PID"
A quick search through this forum found a similar "no such process" error although no remedy was found. Maybe further clues though? Could your pid folder be damaged or odd in some way?
Update: @dvo Oh, forgot to include link to that other thread I mentioned.