nginx日志切割

http://www.ttlsa.com/nginx/nginx-log-cutting/

nginx日志默认情况下统统写入到一个文件中，文件会变的越来越大，非常不方便查看分析。以日期来作为日志的切割是比较好的，通常我们是以每日来做统计的。下面来说说nginx日志切割。
关于nginx相关日志配置参见：《nginx日志配置》一文。logrotate用法参见《logrotate日志管理工具》。

1. 定义日志轮滚策略

# vim nginx-log-rotate

/data/weblogs/*.log {

    nocompress

    daily

    copytruncate

    create

    notifempty

    rotate 7

    olddir /data/weblogs/old_log

    missingok

    dateext

    postrotate

        /bin/kill -HUP `cat /var/run/nginx.pid 2> /dev/null` 2> /dev/null || true

    endscript

}

/data/weblogs/*.log使用通配符时，/data/weblogs/目录下的所有匹配到的日志文件都将切割。如果要切割特定日志文件，就指定到该文件。

2. 设置计划任务

# vim /etc/crontab

59 23 * * * root ( /usr/sbin/logrotate -f /PATH/TO/nginx-log-rotate)

这样每天23点59分钟执行日志切割。

如需转载请注明出处： http://www.ttlsa.com/html/3166.html

在linux环境下，我希望能够每天到0点切换nginx日志，前一天的日志使用日期后缀，并且压缩。

 

步骤：

1.

需要制作一个nginxlogrotate文件，放在目录/etc/logrotate.d/下，

文件内容如下：

/somedir1/*.log {

    daily

    missingok

    rotate 65535

    compress

    dateext

    notifempty

    sharedscripts

    postrotate

        [ ! -f /somedir2/nginx.pid ] || kill -USR1 `cat /somedir2/nginx.pid`

    endscript

}

 

其中somedir1需要替换成存放nginx日志的目录,somedir2需要替换成系统存放nginx.pid文件的目录。

 

2.

定时启动logrotate

设置定时任务，crontab -e

0 0 * * * nohup /usr/sbin/logrotate /etc/logrotate.conf > /dev/null 2>&1 &

 

3.隔天查看nginx日志目录，发现生成了如下的文件

access.log-20121008.gz

error.log-20121008.gz

参见：

http://drumcoder.co.uk/blog/2012/feb/03/nginx-and-logrotate/

 

 

 

 

 

http://drumcoder.co.uk/blog/2012/feb/03/nginx-and-logrotate/

nginx and Logrotate

February 3, 2012

I wanted to start cycling and compressing my previous nginx log files, as they were getting a little large. NGINX doesn't have built in support for log rotation, there is a system level linux command called logrotate that handles this on behalf of many applications. I'm doing this on a Debian squeeze.

/etc/logrotate.d

In /etc/logrotate.d you'll find a file for each application that is using logrotate. There should be one in here called nginx that contains the following:

/var/log/nginx/*.log {
    daily
    missingok
    rotate 52
    compress
    delaycompress
    notifempty
    create 640 root adm
    sharedscripts
    postrotate
            [ ! -f /var/run/nginx.pid ] || kill -USR1 `cat /var/run/nginx.pid`
    endscript
}

This is the default rotate process for nginx files. I store my logs in /home/drumcoder/logs, so we will specify another block in this same file, using many of the same options.

/home/drumcoder/log/*.access.log {
    daily
    missingok
    rotate 52
    mail drumcoder@here.com
    compress
    delaycompress
    notifempty
    sharedscripts
    postrotate
            [ ! -f /var/run/nginx.pid ] || kill -USR1 `cat /var/run/nginx.pid`
    endscript
}

The following options are used:

• daily - rotates the log files daily
• missingok - it's fine if there is no log to rotate
• rotate 52 - rotate logs 52 times, and then delete the oldest
• mail drumcoder@here.com - before you delete an old log, send it to this mail address
• compress - use gzip on old logs
• delaycompress - leave one rotated log that isn't compressed so the process can still write to it if needed
• notifempty - don't rotate empty file
• sharedscripts - when telling nginx that logs have been rotated, only do it once rather than once for each file group

The bottom postrotate block sends a signal to nginx to tell it that logs have been rotated and it should pick up the file handles for the new ones.

References

• https://wincent.com/wiki/nginx_log_rotation
• http://linuxcommand.org/man_pages/logrotate8.html