I had this problem and followed a similar tack to those outlined here.
I had had certbot install a certificate, but it was in certonly --nginx mode, I supplied my own nginx serverblocks. certbot worked, but an nginx failure cast doubt on the accuracy of my provisioning.
This certbot call "restarts" nginx with a modified server block configuration, so it can answer the HTTP-01 challenges. I know this because when it fails, it will log, "nginx restart failed:" just before the bind() failures I'm about to show. My nginx server was down when provisioning succeeded.
I couldn't get systemctl or service to start it and systemd status nginx would only ever show "failed".
Whilst I could get nginx up, and serving, with nginx -s reload I wanted systemd to manage it for me.
No amount of systemctl {start|restart|stop|quit} nginx, would work. The status remained as failed and would show errors with bind():
Oct 07 10:04:13 HostXYZ systemd[1]: Starting A high performance web server and a reverse proxy server...
Oct 07 10:04:13 HostXYZ nginx[17096]: nginx: [emerg] bind() to 0.0.0.0:80 failed (98: Unknown error)
Oct 07 10:04:13 HostXYZ nginx[17096]: nginx: [emerg] bind() to [::]:80 failed (98: Unknown error)
Oct 07 10:04:13 HostXYZ nginx[17096]: nginx: [emerg] bind() to [::]:443 failed (98: Unknown error)
Oct 07 10:04:13 HostXYZ nginx[17096]: nginx: [emerg] bind() to 0.0.0.0:443 failed (98: Unknown error)
That would repeat in journalctl output, 4 or 5 times.
I checked the process and saw:
:~$ ps aux | grep nginx
root 12960 0.0 0.6 77216 9816 ? Ss Oct06 0:00 nginx: master process nginx -c /etc/nginx/nginx.conf
www-data 16944 0.0 0.5 77360 8604 ? S 08:43 0:00 nginx: worker process
That process, which appeared to be occupying the ports needed by my systemd service. My systemd service doesn't use that -c /etc/nginx/nginx.conf. It uses:
ExecStart=/usr/sbin/nginx -g 'daemon on; master_process on;'
nginx -s stop, and quit would not rid me of the rogue process. Instead they both gave the error the OP had:
:~$ sudo nginx -s stop
nginx: [error] open() "/run/nginx.pid" failed (2: No such file or directory)
Both my systemd service unit and /etc/nginx/nginx.conf gave /run/nginx.pid as the PIDFile/pid. For some reason, /etc/nginx/nginx.conf wasn't creating it.
What I needed to do:
sudo killall nginx
sudo systemctl start nginx
That knocked out the other nginx service (I think it came from nginx -s reload but I couldn't shut it down by the corollary command) Which looked like this:
:~$ sudo killall nginx
:~$ ps aux | grep nginx
john 17140 0.0 0.1 4008 2004 pts/0 S+ 10:10 0:00 grep --color=auto nginx
:~$ sudo systemctl start nginx
:~$ sudo systemctl status nginx
● nginx.service - A high performance web server and a reverse proxy server
Loaded: loaded (/lib/systemd/system/nginx.service; enabled; vendor preset: enabled)
Active: active (running) since Fri 2022-10-07 10:10:25 UTC; 1s ago
...
:~$ ps aux | grep nginx
root 11481 0.0 0.1 76484 2588 ? Ss 10:10 0:00 nginx: master process /usr/sbin/nginx -g daemon on; master_process on;
www-data 11482 0.0 0.2 76876 4284 ? S 10:10 0:00 nginx: worker process
:~$ cat /run/nginx.pid
11481
