Hello Roberto. On 07 Jul 2014, at 10:33, Roberto De Ioris <[email protected]> wrote:
> >> Hello Roberto. >> >> On 01 Jul 2014, at 08:18, Roberto De Ioris <[email protected]> wrote: >> >>> If i remember correctly you have an https router on front, could it be >>> it >>> is the guilty one ? (there is no graceful reload concept in gateways, >>> and >>> when you reload the master, gateways are destroyed asap) >> >> Yes, you are right. So there is no way to add a cron entry and gracefully >> reload in this case, correct? > > probably i would split the https router from the app-instance. (or if you > have a proxy/load-balancer on front of your https router, consider > removing it and directly configure uWSGI to speak https with > --https-socket :port,cert,key) I don't have a proxy/LB in front of uWSGI. I use round-robin DNS and Corosync for when a node goes down. In this case if I use https-socket directly graceful reload will work (as in current requests will finish), but new requests will not be served until all workers recycle, right? If I separate the https router won't I have the same problem? If the app-instance is waiting for all the workers to gracefully shutdown, it will not be servicing requests coming in from the router. Are you suggesting multiple app-instances? Best regards, André
signature.asc
Description: Message signed with OpenPGP using GPGMail
_______________________________________________ uWSGI mailing list [email protected] http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi
