Ubuntu Apache-随机中断

最近我遇到了Apache的新问题。我们有一个在Flask（1.0.2）中运行的Python（3.5）应用程序

$ lsb_release -a
No LSB modules are available.
Distributor ID: Ubuntu
Description:    Ubuntu 16.04.2 LTS
Release:    16.04
Codename:   xenial

我们在ELB（AWS Elasict Loadbalancer）后面有两台服务器，它们突然（在过去3个月的配置上运行）开始出现故障。我发现通过ELB的警报和外部监控工具。我们突然开始得到E408（超时）和E503（服务不可用）

我开始挖掘，看看是什么原因， 在apache日志中，我发现许多消息（似乎就在失败之前）如下：

[Mon Jun 25 22:27:04.613967 2018] [wsgi:error] [pid 1275:tid 139684390848256] (70008)Partial results are valid but processing is incomplete: [client 1.2.3.4:2819] mod_wsgi (pid=1275): Unable to get bucket brigade for request., referer: https://xx.xx.xx/

Jun 25 22:13:25 my_hostname systemd[1]: Created slice User Slice of ubuntu.
Jun 25 22:13:25 my_hostname systemd[1]: Starting User Manager for UID 1000...
Jun 25 22:13:25 my_hostname systemd[1]: Started Session 1424 of user ubuntu.
Jun 25 22:13:25 my_hostname systemd[6239]: Reached target Sockets.
Jun 25 22:13:25 my_hostname systemd[6239]: Reached target Timers.
Jun 25 22:13:25 my_hostname systemd[6239]: Reached target Paths.
Jun 25 22:13:25 my_hostname systemd[6239]: Reached target Basic System.
Jun 25 22:13:25 my_hostname systemd[6239]: Reached target Default.
Jun 25 22:13:25 my_hostname systemd[6239]: Startup finished in 8ms.
Jun 25 22:13:25 my_hostname systemd[1]: Started User Manager for UID 1000.
Jun 25 22:14:47 my_hostname systemd[1]: Stopping LSB: Apache2 web server...
Jun 25 22:14:47 my_hostname apache2[6624]:  * Stopping Apache httpd web server apache2
Jun 25 22:14:59 my_hostname apache2[6624]:  *
Jun 25 22:14:59 my_hostname systemd[1]: Stopped LSB: Apache2 web server.
Jun 25 22:14:59 my_hostname systemd[1]: Starting LSB: Apache2 web server...
Jun 25 22:14:59 my_hostname apache2[6660]:  * Starting Apache httpd web server apache2
Jun 25 22:14:59 my_hostname apache2[6660]: AH00557: apache2: apr_sockaddr_info_get() failed for my_hostname
Jun 25 22:14:59 my_hostname apache2[6660]: AH00558: apache2: Could not reliably determine the server's fully qualified domain name, using 127.0.0.1. Set the 'ServerName' directive globally to suppress this message
Jun 25 22:15:00 my_hostname apache2[6660]:  *
Jun 25 22:15:00 my_hostname systemd[1]: Started LSB: Apache2 web server.

我还查询了syslog，看到了这个：

[Mon Jun 25 22:27:04.613967 2018] [wsgi:error] [pid 1275:tid 139684390848256] (70008)Partial results are valid but processing is incomplete: [client 1.2.3.4:2819] mod_wsgi (pid=1275): Unable to get bucket brigade for request., referer: https://xx.xx.xx/

Jun 25 22:13:25 my_hostname systemd[1]: Created slice User Slice of ubuntu.
Jun 25 22:13:25 my_hostname systemd[1]: Starting User Manager for UID 1000...
Jun 25 22:13:25 my_hostname systemd[1]: Started Session 1424 of user ubuntu.
Jun 25 22:13:25 my_hostname systemd[6239]: Reached target Sockets.
Jun 25 22:13:25 my_hostname systemd[6239]: Reached target Timers.
Jun 25 22:13:25 my_hostname systemd[6239]: Reached target Paths.
Jun 25 22:13:25 my_hostname systemd[6239]: Reached target Basic System.
Jun 25 22:13:25 my_hostname systemd[6239]: Reached target Default.
Jun 25 22:13:25 my_hostname systemd[6239]: Startup finished in 8ms.
Jun 25 22:13:25 my_hostname systemd[1]: Started User Manager for UID 1000.
Jun 25 22:14:47 my_hostname systemd[1]: Stopping LSB: Apache2 web server...
Jun 25 22:14:47 my_hostname apache2[6624]:  * Stopping Apache httpd web server apache2
Jun 25 22:14:59 my_hostname apache2[6624]:  *
Jun 25 22:14:59 my_hostname systemd[1]: Stopped LSB: Apache2 web server.
Jun 25 22:14:59 my_hostname systemd[1]: Starting LSB: Apache2 web server...
Jun 25 22:14:59 my_hostname apache2[6660]:  * Starting Apache httpd web server apache2
Jun 25 22:14:59 my_hostname apache2[6660]: AH00557: apache2: apr_sockaddr_info_get() failed for my_hostname
Jun 25 22:14:59 my_hostname apache2[6660]: AH00558: apache2: Could not reliably determine the server's fully qualified domain name, using 127.0.0.1. Set the 'ServerName' directive globally to suppress this message
Jun 25 22:15:00 my_hostname apache2[6660]:  *
Jun 25 22:15:00 my_hostname systemd[1]: Started LSB: Apache2 web server.

有趣的是，两台服务器（几乎完全相同）同时出现故障（由于部署了新版本，它们几乎在同一时间重新启动，而且两台服务器的通信量可能大致相同，因为它们位于一个负载平衡器后面）

已经试图找到类似的问题，但迄今为止运气不佳

还有一件有趣的事，我在日志中发现了一些类似的消息：

[Mon Jun 25 22:27:04.657763 2018] [wsgi:error] [pid 1274:tid 139684507617024] [remote 172.31.12.149:720] mod_wsgi (pid=1274): Exception occurred processing WSGI script '/home/ubuntu/my_app/app.wsgi'.
[Mon Jun 25 22:27:04.658503 2018] [wsgi:error] [pid 1274:tid 139684482414336] [remote 172.31.12.149:62417] mod_wsgi (pid=1274): Exception occurred processing WSGI script '/home/ubuntu/my_app/app.wsgi'.
[Mon Jun 25 22:27:04.658528 2018] [wsgi:error] [pid 1274:tid 139684532819712] [remote 172.31.12.149:52139] mod_wsgi (pid=1274): Exception occurred processing WSGI script '/home/ubuntu/my_app/app.wsgi'.
[Mon Jun 25 22:27:04.658584 2018] [wsgi:error] [pid 1274:tid 139684482414336] [remote 172.31.12.149:62417] OSError: failed to write data
[Mon Jun 25 22:27:04.658818 2018] [wsgi:error] [pid 1274:tid 139684516017920] [remote 172.31.12.149:208] OSError: failed to write data
[Mon Jun 25 22:27:04.659999 2018] [wsgi:error] [pid 1274:tid 139684532819712] [remote 172.31.12.149:52139] OSError: failed to write data
[Mon Jun 25 22:27:04.660411 2018] [wsgi:error] [pid 1274:tid 139684507617024] [remote 172.31.12.149:720] OSError: failed to write data

不确定是否相关，但我确定我们在完成很多请求之前取消了它们（出于某种原因）

此外，我们在Ubuntu+Flask上运行了多年（很可能是相同的设置），我们从未遇到过这样的问题

---

非常感谢您的任何建议，谢谢

您使用的是什么mod_wsgi配置？您是否使用mod_wsgi的守护程序模式并设置所需的超时，以便在发生锁定时自动恢复？您是否通过将

WSGIApplicationGroup

设置为

%{GLOBAL}

来强制使用主Python解释器，以避免因Python的第三方C扩展而导致无法在子解释器中工作的锁定。确实需要查看您的配置。@GrahamDumpleton我们正面临同样的问题，通过将

WSGIApplicationGroup

设置为

%{GLOBAL}