从Apache（mpm prefork）转换到php5-fpm的Nginx，我发现错误日志不愉快的parsing。
例如Nginx会产生这样的东西（具体的编辑）：
2012/03/12 09:56:59 [error] 22075#0: *432 FastCGI sent in stderr: "PHP Warning: require_once([somefile]): failed to open stream: No such file or directory in [sourcefile] on line 6 PHP Fatal error: require_once(): Failed opening required '[somefile]' (include_path='.:/usr/share/php:/usr/share/pear') in [sourcefile] on line 6" while reading response header from upstream, client: [ip], server: [host].*, request: "GET / HTTP/1.1", upstream: "fastcgi://127.0.0.1:9000", host: "[host]" 其中包含了大量的信息，我没有真正感兴趣的99％的时间，而开发/debugging。
Apache用来输出更可读的东西，比如：
[date] [error] [client] PHP Warning: require_once([somefile]): failed to open stream: No such file or directory in [sourcefile] on line 6 [date] [error] [client] PHP Fatal error: require_once(): Failed opening required '[somefile]' (include_path='.:/usr/share/php:/usr/share/pear') in [sourcefile] on line 6"
由于实际的错误首先出现（与FPM细节相对）并且在terminal中列alignment，因此大多更容易扫描。
有什么办法可以让Nginx的行为更像Apache的错误日志？