來源：http://www.aisir.cn/2017/07/15/aws實例部署flask報錯script-timed-out-before-returning-headers-application-py/

[Thu Jul 13 13:21:32.542375 2017] [core:error] [pid 25744] [client 127.0.0.1:48852] Script timed out before returning headers: application.py
[Thu Jul 13 13:21:32.618324 2017] [core:error] [pid 25672] [client 127.0.0.1:48754] Script timed out before returning headers: application.py
[Thu Jul 13 13:21:33.517055 2017] [core:error] [pid 25702] [client 127.0.0.1:48846] Script timed out before returning headers: application.py
[Thu Jul 13 13:21:33.593206 2017] [core:error] [pid 25722] [client 127.0.0.1:48814] Script timed out before returning headers: application.py

這是一個被困擾需求的問題，接收SQS發來的消息總量有一萬多條，並不是很多，一直以來是以為flask框架的問題，所以解決方法一直是

產生的背景：

抓取Facebook廣告數據並通過aws的SQS分發處理，SQS數量在萬級，同時提取廣告數據中的圖片視頻信息，執行時間多消耗在網絡請求。

問題解決：

配置機器實例

可見性超時（SQS visibility timeout）：3600

官方解釋：在將入站消息返回隊列之前將其鎖定以供處理的秒數。明白的說就是大於應用處理SQS消息的時間設置

不活動超時（inactivity timeout）:1800

官方解釋 ：在現有連接上等待來自應用程序的響應的秒數。即等多久就報time out錯誤，也就是上面看到的報錯。

註意：不活動超時需要小於可見性超時，否則報錯 The inactivity timeout you specified for your worker tier environment is greater than the SQS visibility timeout. Be aware that this can result in multiple retries, which can increase the resources your application tasks consume. Consider an inactivity timeout that is not greater than 300.

Stack Overflow上對於該問題的解決方式是，在.ebextensions中命令，測試無效

:files:
"/etc/httpd/conf.d/wsgi_custom.conf":
mode: "000644"
owner: root
group: root
content: |
WSGIApplicationGroup %{GLOBAL}

aws實例部署flask報錯script-timed-out-before-returning-headers-application-py