I've discovered something simillar lately (somewhere between 0.9.7.2 and
0.9.8, haven't tested in on the official 0.9.8, only commits in between)
when using uwsgi in --lazy mode (single process, no emperor, just one
mgmt process, py2.6). When uwsgi tries to create an application instance
and that fails (for ex. because of invalid import), uwsgi keeps retrying
but it constantly fails so an app is never created. But then if I try to
end uwsgi process (sinal 3) it locks itself in infinite loop so KILL is
necessary.
I can provide more details if needed.
W dniu 20.06.2011 03:52, Trey Long pisze:
From a previous conversation I learned that TOUCH_RELOAD could be
passed as UWSGI variable. My current setup:
server {
listen 8080;
server_name ...;
location / {
include uwsgi_params;
uwsgi_pass unix:///var/tmp/uwsgi.sock;
uwsgi_param UWSGI_CHDIR /appdir/;
uwsgi_param UWSGI_SCRIPT app.wsgi;
uwsgi_param UWSGI_TOUCH_RELOAD /appdir/app.wsgi;
}
location /static/ {
root /appdir/static;
}
}
This works fine, when I touch the file it does update. However, uWSGI
continuously spawns new interpreters to handle the same application.
It appears like the original mod time isn't updated when a new mod
time is detected and every subsequent comparison yields a new
interpreter instance:
"WSGI application 10 (SCRIPT_NAME=ubuntu.local:8080|) ready on
interpreter 0x5868e90 pid: 15701 (default app)"
Keep in mind I only have a single app in this setup.
Trey
_______________________________________________
uWSGI mailing list
[email protected]
http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi
--
Pozdrawiam,
Łukasz Czuja
Programista / Analityk / Architekt IT
_______________________________________________
uWSGI mailing list
[email protected]
http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi
