使用 Rsyslog 将 Nginx Access Log 写入 Kafka

环境说明

  • CentOS Linux release 7.3.1611
  • kafka_2.12-0.10.2.2
  • nginx/1.12.2
  • rsyslog-8.24.0-34.el7.x86_64.rpm

创建测试 Topic

$ ./kafka-topics.sh --zookeeper 192.168.72.25:2181/kafka --create --topic develop-test-topic --partitions 10 --replication-factor 3

Rsyslog 安装

一般系统自带 Rsyslog 服务无需另外安装。但是因为数据需要通过 Rsyslog 的 omkafka 模块写入到 Kafka,而 omkafka 在 Rsyslog 的 v8.7.0+ 版本才支持,所以要看当前系统中 Rsyslog 的版本是否需要升级。使用以下命令查看:

# rsyslogd -v
rsyslogd 7.4.7, compiled with:
    FEATURE_REGEXP:				Yes
    FEATURE_LARGEFILE:			No
    GSSAPI Kerberos 5 support:		Yes
    FEATURE_DEBUG (debug build, slow code):	No
    32bit Atomic operations supported:	Yes
    64bit Atomic operations supported:	Yes
    Runtime Instrumentation (slow code):	No
    uuid support:				Yes

See http://www.rsyslog.com for more information.

执行以下命令安装:

# sudo yum install rsyslog

安装依赖关系如下:

添加 omkafka 模块

# yum install rsyslog-kafka

Rsyslog Client Nginx 配置

注意,Nginx 1.7.1 之后才支持 syslog 的方式处理日志。具体配置项参见官网文档Logging to syslog

Nginx 配置主要是日志格式和 Access Log 配置项:

log_format  main  '$remote_addr - $remote_user [$time_local] "$request" '
                  '$status $body_bytes_sent "$http_referer" '
                  '"$http_user_agent" "$http_x_forwarded_for"';

server {
    listen 11000;

    location / {
        proxy_pass http://10.16.0.144:11000;
        access_log syslog:server=localhost,facility=local7,tag=nginx11000Root,severity=info main;
    }
}

Rsyslog Server 端配置

Rsyslog 的主配置文件 /etc/rsyslog.conf,其中会包含引入 /etc/rsyslog.d 下扩展名为 conf 的配置文件。

修改配置文件 /etc/rsyslog.conf 将下面两行前面的注释去掉:

$ModLoad imudp
$UDPServerRun 514

在 /etc/rsyslog.d 目录下创建 rsyslog_nginx_kafka_cluster.conf,配置内容如下:

module(load="imudp")
input(type="imudp" port="514")

# nginx access log ==> rsyslog server(local) ==> kafka
module(load="omkafka")

template(name="nginx-11000-root" type="string" string="%msg%")

if $inputname == "imudp" then {
    if ($programname == "nginx11000Root") then
        action(type="omkafka"
            template="nginx-11000-root"
            broker=["192.168.72.10:9092","192.168.72.20:9092","192.168.72.25:9092","192.168.72.26:9092","192.168.72.27:9092","192.168.72.48:9092","192.168.72.55:9092","192.168.72.80:9092","192.168.72.81:9092","192.168.72.97:9092"]
            topic="develop-test-topic"
            partitions.auto="on"
            confParam=[
                "socket.keepalive.enable=true"
            ]
        )
}

:rawmsg, contains, "nginx11000Root" ~

联调测试

启动 Rsyslog 服务:

# service rsyslog start
Redirecting to /bin/systemctl start  rsyslog.service

遇到的问题

syslog tag 只能包含字母和数字

# nginx -t
nginx: [emerg] syslog "tag" only allows alphanumeric characters and underscore in     /etc/nginx/conf.d/jx-11000-jenkins149-36-144.conf:7
nginx: configuration file /etc/nginx/nginx.conf test failed

‘omkafka’ is unknown

Rsyslog 中没有包含 omkafka 模块,需要另外安装。查看 /var/log/messages 日志信息会有以下提示:

# tail -f messages
Mar 15 15:13:40 192-168-72-29 systemd: Started System Logging Service.
Mar 15 15:13:40 192-168-72-29 rsyslogd: could not load module '/usr/lib64/rsyslog/omkafka.so', dlopen:     /usr/lib64/rsyslog/omkafka.so: cannot open shared object file: No such file or directory  [v8.24.0-34.el7 try     http://www.rsyslog.com/e/2066 ]
Mar 15 15:13:40 192-168-72-29 rsyslogd: could not load module '/usr/lib64/rsyslog/omkafka.so', dlopen:     /usr/lib64/rsyslog/omkafka.so: cannot open shared object file: No such file or directory  [v8.24.0-34.el7 try     http://www.rsyslog.com/e/2066 ]
Mar 15 15:13:40 192-168-72-29 rsyslogd: module name 'omkafka' is unknown [v8.24.0-34.el7 try     http://www.rsyslog.com/e/2209 ]
Mar 15 15:13:40 192-168-72-29 rsyslogd: module name 'omkafka' is unknown [v8.24.0-34.el7 try     http://www.rsyslog.com/e/2209 ]

CentOS 6.5 升级 Rsyslog

CentOS 6.5 自带的 Rsyslog 版本是 rsyslogd 5.8.10。按照以下方式安装新版本:

# cd /etc/yum.repos.d/
# wget http://rpms.adiscon.com/v8-stable/rsyslog.repo
# yum install rsyslog