Open KES777 opened 9 years ago
Do you have some idea on how to accomplish this ? uWSGI configuration system is not linear (it is more like a language). The --fallback-config option seems currently the safest and easiest way to reduce this kind of problems (albeit not perfect)
run second instance of uwsgi before killing master. The second instance do all the same as normal start except do not bind to port and do not write any log messages. Then you check exit code. If it is OK then you can kill current and start new.
@KES777 well at this point --fallback-config gives you the same reliability without the need for generating a whole new stack and changing the code paths (that is an hard task)
When we restart server and configuration is not OK the server will down until we fix configuration. That is wrong behaviour and does not follow zero time down. Detail info here
You can see that server is going down and do not up...