Nginx上通过uWSGI发出的烧瓶和被信号9杀死的工作者

Question

由于某些原因，当我的Flas应用程序部署到Docker时，它将在多个请求上崩溃并出现错误

nginx    | 2020/06/26 20:41:39 [error] 20#20: *1 upstream prematurely closed connection while reading response header from upstream, client: 172.18.0.1, server: , request: "POST /api/v1.0/export HTTP/1.1", upstream: "uwsgi://172.18.0.2:5555", host: "127.0.0.1:80"
flask    | DAMN ! worker 4 (pid: 16) died, killed by signal 9 :( trying respawn ...
flask    | Respawned uWSGI worker 4 (new pid: 18)

在venv上运行时，该应用可以正常运行

uWSGI配置如下：

[uwsgi]
wsgi-file = run.py
callable = app
socket = :5555
processes = 16
threads = 2
master = true
chmod-socket = 660
vacuum = true
die-on-term = true
harakiri-verbose = true

和nginx的配置如下：

server {

    listen 80;

    location / {
        include uwsgi_params;
        uwsgi_pass flask:5555;
        uwsgi_read_timeout 300;
    }

    client_max_body_size 100M;

    client_header_timeout       120s;
    client_body_timeout         120s;
    keepalive_timeout           120s;
    send_timeout                120s;
}

对此有何想法？