You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Originally posted by zeghong January 18, 2022
I was wondering if we can add an option to customize the upstream_keepalive_idle_timeout per upstream object.
There is a global configuration upstream_keepalive_idle_timeout (Link) which controls the global default upstream keepalive timeout, but i think it would be helpful if we can customize this value per upstream object.
Here is a typical scenario: Gunicorn is a popular Python WSGI HTTP server, and it's basic sync worker does not support persistent connections. So the deployment guide recommends the nginx directive keepliave_timeout been set to 5 (seconds), which is shorter than kong's default value 60。It the keepalive idle timeout can be customized without affecting other workload with persistent connections support, it will helps to reduce the number of TIME_WAIT TCP sockets.
Since the original nginx keepalive_timeoutdirective value is set in the upstream context, i assumes this feature is realizable from technical point of view.
The text was updated successfully, but these errors were encountered:
Discussed in #8302
Originally posted by zeghong January 18, 2022
I was wondering if we can add an option to customize the
upstream_keepalive_idle_timeout
per upstream object.There is a global configuration
upstream_keepalive_idle_timeout
(Link) which controls the global default upstream keepalive timeout, but i think it would be helpful if we can customize this value per upstream object.Here is a typical scenario:
Gunicorn is a popular Python WSGI HTTP server, and it's basic sync worker does not support persistent connections. So the deployment guide recommends the nginx directive
keepliave_timeout
been set to5
(seconds), which is shorter than kong's default value60
。It the keepalive idle timeout can be customized without affecting other workload with persistent connections support, it will helps to reduce the number ofTIME_WAIT
TCP sockets.Since the original nginx
keepalive_timeout
directive value is set in the upstream context, i assumes this feature is realizable from technical point of view.The text was updated successfully, but these errors were encountered: