feat: As a user, I want to set enable_cpu_affinity default to false, so that it won't affect running APISIX via container

[spacewander]

Description

Nowadays, more and more users deploy APISIX via container. As Nginx's worker_cpu_affinity doesn't count with cgroup, enabling worker_cpu_affinity by default will affect the behavior of APISIX, for instance, multiple instances will be bound to one CPU. To avoid this problem, it would be better to disable CPU affinity option by default in the conf/config-default.yaml.

The worker_processes option also has a similar problem, but changing this option will bring enormous breaking change.

[kellyseeme]

how worker_processes set in k8s enviroment? if there set the cpu and mem request and limit.

[tokers]

how worker_processes set in k8s enviroment? if there set the cpu and mem request and limit.

The default value of worker_processes is auto (see https://github.com/apache/apisix-helm-chart/blob/master/charts/apisix/values.yaml#L214) if you install APISIX via helm chart.

[tokers]

how worker_processes set in k8s enviroment? if there set the cpu and mem request and limit.

The default value of worker_processes is auto (see https://github.com/apache/apisix-helm-chart/blob/master/charts/apisix/values.yaml#L214) if you install APISIX via helm chart.

The number of APISIX worker processes is not related to the CPU / memory request & limitation. By the way, this is off the topic.

[Hazel6869]

Hi, can you assign the issue to me ? @spacewander

[kellyseeme]

@tokers thanks for reply,maybe the woker_processes is the relevent topic.if we set auto,this worker_processes is set to the node of the number cpus,maybe this is not resonable。

In kubernetes, nginx spawn worker processes depending on the resource request for the pod by default. e.g if you have the following in your deployment:

resources:
requests:
memory: 2048Mi
cpu: 2000m
Then nginx will spawn 2 worker process (2000 milli cpu = 2 cpu)

[tokers]

In kubernetes, nginx spawn worker processes depending on the resource request for the pod by default. e.g if you have the following in your deployment:

Any docs about it?