Styjun

How did Apollo 15's depressurization work?

Does C++20 mandate source code being stored in files?

Is "stainless" a bulk or a surface property of stainless steel?

I think my coworker went through my notebook and took my project ideas

Metal that glows when near pieces of itself

What professions does medieval village with a population of 100 need?

Hai la patente? - omission of possessive adjective

Nuclear decay triggers

Is Eldritch Blast affected by antimagic field

Chess software to analyze games

Alchemist potion on Undead

Do living authors still get paid royalties for their old work?

Are there categories whose internal hom is somewhat 'exotic'?

Why doesn't the Falcon-9 first stage use three legs to land?

!I!n!s!e!r!t! !n!b!e!t!w!e!e!n!

What is the latest version of SQL Server native client that is compatible with Sql Server 2008 r2

How to compare two different formulations of a problem?

Are illustrations in novels frowned upon?

Count the frequency of items in an array

Did the twin engined Lazair ultralight have a throttle for each engine?

Why don't sharp and flat root note chords seem to be present in much guitar music?

To "hit home" in German

How can I describe being temporarily stupid?

Why does my air conditioner still run, even when it is cooler outside than in?

Reloading gracefully with Zerg pool isn't really graceful while workers load

Bad Django / uwsgi performanceuwsgi mapping request to the processCall python code from uWSGI master cycleuWSGI: Using emperor mode, register internal routing rules for vassalsuwsgi working with nginx, systemd emperor not working with same configs and setupHow to avoid Django sub thread being killed by uWSGI respawnuWSGI restarts attached processes without stopping old onesuwsgi with emperor/systemd: announcing loyalty takes up to 3 minutesHow to fix uWSGI internal routing?First request doesnt terminate (no FIN) uWSGI + nginx

.everyoneloves__top-leaderboard:empty,.everyoneloves__mid-leaderboard:empty,.everyoneloves__bot-mid-leaderboard:empty margin-bottom:0;

0

I am using an emperor to load a zerg pool, in order to gracefuly deploy different versions. I'm using this method:
http://zarnovican.github.io/2016/02/15/uwsgi-graceful-reload/#zerg-dance_1
My main problem is that my workers need to build an internal cache before actually accepting requests, and this can take up to 20 seconds per worker.

This means that when they are started socket opens - they still can't handle requests. So now when I try to switch versions, some requests arrive to to starting workers and receive timeouts there.
Is there a way to either:
- notify uwsgi they are not ready yet to receive requests until they say they do.
- or - start with a different socket and then when ready reasign to the main socket?

This is on a Linux server, with python 3.6.

uwsgi

share|improve this question

asked Mar 27 at 14:49

Yoav MoranYoav Moran

1433 bronze badges

add a comment | 

0

I am using an emperor to load a zerg pool, in order to gracefuly deploy different versions. I'm using this method:
http://zarnovican.github.io/2016/02/15/uwsgi-graceful-reload/#zerg-dance_1
My main problem is that my workers need to build an internal cache before actually accepting requests, and this can take up to 20 seconds per worker.

This means that when they are started socket opens - they still can't handle requests. So now when I try to switch versions, some requests arrive to to starting workers and receive timeouts there.
Is there a way to either:
- notify uwsgi they are not ready yet to receive requests until they say they do.
- or - start with a different socket and then when ready reasign to the main socket?

This is on a Linux server, with python 3.6.

uwsgi

share|improve this question

asked Mar 27 at 14:49

Yoav MoranYoav Moran

1433 bronze badges

add a comment | 

I am using an emperor to load a zerg pool, in order to gracefuly deploy different versions. I'm using this method:
http://zarnovican.github.io/2016/02/15/uwsgi-graceful-reload/#zerg-dance_1
My main problem is that my workers need to build an internal cache before actually accepting requests, and this can take up to 20 seconds per worker.

This means that when they are started socket opens - they still can't handle requests. So now when I try to switch versions, some requests arrive to to starting workers and receive timeouts there.
Is there a way to either:
- notify uwsgi they are not ready yet to receive requests until they say they do.
- or - start with a different socket and then when ready reasign to the main socket?

This is on a Linux server, with python 3.6.

uwsgi

share|improve this question

asked Mar 27 at 14:49

Yoav MoranYoav Moran

1433 bronze badges

share|improve this question

asked Mar 27 at 14:49

Yoav MoranYoav Moran

1433 bronze badges

share|improve this question

asked Mar 27 at 14:49

Yoav MoranYoav Moran

1433 bronze badges

asked Mar 27 at 14:49

Yoav MoranYoav Moran

1433 bronze badges

asked Mar 27 at 14:49

Yoav MoranYoav Moran

1433 bronze badges