When uWSGI start via Upstart I have independent Emperor instances whith two process each (as manually run, see below).
Upstart create multiple emperor processes:
# ps ax | grep -i emper
888 ? S 0:53 uwsgi /etc/uwsgi/emperor.ini
896 ? S 0:33 uwsgi /etc/uwsgi/emperor.ini
904 ? S 0:52 uwsgi /etc/uwsgi/emperor.ini
905 ? S 0:34 uwsgi /etc/uwsgi/emperor.ini
...
When I kill all Emperors and run it manually I have two Emperor's processes with workers for sites:
# pstree -up | less
...
-uwsgi(16347)---uwsgi(16348)-+-uwsgi(16349,web)-+-uwsgi(16358)
| | |-uwsgi(16360)
| | |-uwsgi(17058)
| | |-uwsgi(17261)
| | |-uwsgi(17410)
| | |-uwsgi(17488)
| | |-uwsgi(17750)
| | `-uwsgi(17770)
| `-uwsgi(16350,web)-+-uwsgi(16361)
| |-uwsgi(16362)
| |-uwsgi(16363)
| |-uwsgi(16364)
| |-uwsgi(16365)
| |-uwsgi(16366)
| |-uwsgi(16367)
| `-uwsgi(16368)
My Upstart config for Emperor:
# Emperor uWSGI script
description "uWSGI Emperor"
start on runlevel [2345]
stop on runlevel [06]
respawn
exec uwsgi /etc/uwsgi/emperor.ini
My uWSGI Emperor config:
[uwsgi]
emperor-tyrant = true
emperor = %dvassals
emperor-pidfile = %demperor.pid
emperor-stats = %demperor.stat.sock
vassals-include = %dvassals-default.ini
touch-logrotate = %p
touch-reload = %p
touch-reload = %dvassals-default.ini
log-date = true
log-truncate = true
daemonize = /var/log/uwsgi/emperor.log
May be I wrong on Upstart config? Logs didn't help me.
Need change daemonize to logto.
More info in issue in uWSGI repo: https://github.com/unbit/uwsgi/issues/1510