mariadb审计日志导入 hive

Posted 2020-12-25 运维从入门到放弃

我们使用的 mariadb, 用的这个审计工具 https://mariadb.com/kb/en/library/mariadb-audit-plugin/

这个工具一点都不考虑后期对数据的处理, 因为他的日志是这样的

20180727 11:40:17,aaa-main-mariadb-bjc-001,user,10.1.111.11,3125928,6493942844,QUERY,account,\'select


    id, company_id, user_id, department, title, role, create_time, update_time, status,
    is_del, receive_email, contact

    from company


     WHERE (  user_id = 101



                  and is_del = 0 )\',0　　

所以需要上 logstash 格式化一下

input {
  file {
    path => ["/data/logs/mariadb/server_audit.log"]
    start_position => "end"
    codec => multiline {
      charset => "ISO-8859-1"
      pattern => "^[0-9]{8}"
      negate => true
      what => "previous"
    }
  }
}

filter {
    if "quartz" in [message] { drop {} }
    mutate {
        gsub => [
            "message", "\\s", " ",
            "message", "\\s+", " "
        ]
    }
    dissect {
        mapping => {
            "message" => "%{ts} %{+ts},%{hostname},%{user},%{dbhost},%{connid},%{queryid},%{operate},%{db},%{object}"
        }
    }
    mutate {
        replace => { "message" => "%{ts}    %{hostname}    %{user}    %{dbhost}    %{operate}    %{db}    %{object}" }
    }
}

output {
    file {
        path => "/data/logs/mariadb/%{host}_%{+YYYY-MM-dd_HH}.gz"
        gzip => true
        codec => line { format => "%{message}" }
    }
}

注意 !!!

replace 那个地方各个字段之间我用的 tab 隔开的, 如果用 vim 一定不能写\\t, 这在hive中不识别的, 要在 vim 中先按 ctrl+v, 再按 tab

在 vim 中 set list 如下显示才对

mariadb 的审计日志不能按小时切割,上面 logstash 我把日志按小时生成 gz 文件了,后面就是推到 hdfs 中了, 期间试了各种方法

logstash的 output [webhdfs] 效率不行还丢数据

syslog-ng

rsyslog

统统不好用,最终直接使用 hdfs cli简直完美

我把 logstash 的配置和推送到 hdfs 的命令都打到 rpm 里,下面贴一下 rpm SPEC 文件 也记录一下吧

Name:           logstash
Version:        1.0.0
Release:        1%{?dist}
Summary:        specialize to mysql audit log collection
License:        GPL
AutoReqProv:    no

%define __os_install_post %{nil}

%description

%prep

%build

%install
rm -rf $RPM_BUILD_ROOT
mkdir -p %{buildroot}/{apps,etc,usr,var/lib/logstash,var/log/logstash}
cp -r %{_builddir}/etc/* %{buildroot}/etc/
cp -r %{_builddir}/usr/* %{buildroot}/usr/
cp -r %{_builddir}/apps/* %{buildroot}/apps/

%post
chown -R root:root /usr/share/logstash
chown -R root /var/log/logstash
chown -R root:root /var/lib/logstash
chown -R root:root /apps/hadoop-2.6.0
/usr/share/logstash/bin/system-install /etc/logstash/startup.options
cat >> /etc/hosts <<EOF
# for logstash push msyql audit to HDFS
这里填上 hdfs namenode和 datanode 的 hosts
EOF

echo "$(shuf -i 3-15 -n 1) * * * *" \'source /etc/profile;/apps/hadoop-2.6.0/bin/hdfs dfs -copyFromLocal /data/logs/mariadb/${HOSTNAME}_$(date -u +"\\%Y-\\%m-\\%d_\\%H" -d "last hour").gz hdfs://active_namenode/mysql_audit/$(date -u +"\\%Y-\\%m-\\%d")/ && rm -f /data/logs/mariadb/${HOSTNAME}_$(date -u +"\\%Y-\\%m-\\%d_\\%H" -d "last hour").gz\' >> /var/spool/cron/root

initctl start logstash

%files
%defattr(-,root,root)
/apps/hadoop-2.6.0
/etc/logstash
/usr/share/logstash
/var/lib/logstash
/var/log/logstash


%preun
if [ $1 -eq 0 ]; then
  # Upstart
  if [ -r "/etc/init/logstash.conf" ]; then
    if [ -f "/sbin/stop" ]; then
      /sbin/stop logstash >/dev/null 2>&1 || true
    else
      /sbin/service logstash stop >/dev/null 2>&1 || true
    fi
    if [ -f "/etc/init/logstash.conf" ]; then
      rm /etc/init/logstash.conf
    fi
  # SYSV
  elif [ -r "/etc/init.d/logstash" ]; then
    /sbin/chkconfig --del logstash
    if [ -f "/etc/init.d/logstash" ]; then
      rm /etc/init.d/logstash
    fi
  # systemd
  else
    systemctl stop logstash >/dev/null 2>&1 || true
    if [ -f "/etc/systemd/system/logstash-prestart.sh" ]; then
      rm /etc/systemd/system/logstash-prestart.sh
    fi

    if [ -f "/etc/systemd/system/logstash.service" ]; then
      rm /etc/systemd/system/logstash.service
    fi
  fi
  if getent passwd logstash >/dev/null ; then
    userdel logstash
  fi

  if getent group logstash > /dev/null ; then
    groupdel logstash
  fi
fi


%postun


%clean
rm -rf $RPM_BUILD_ROOT

我把 hadoop 的程序也放进去了,方便

安装完 rpm 自动启动 logstash   省劲

现在日志已经按天写到 hdfs 中,下面再导入 hive 中

先创建 hive 表

create table mysql_audit(datetime string,hostname string,username string,dbhost string,operation string,db string,object string) partitioned by (dt int,hour smallint,module string) row format delimited fields terminated by \'\\t\';

分了3个 partition

load hdfs to hive

#!/bin/bash
# Description: load hdfs mysql audit gz to hive
# Author     : quke
# Date       : 2018-07-27

source /root/.bash_profile

cur_date=$(date -u +"%Y-%m-%d" -d "last hour")
cur_date_short=$(date -u +"%Y%m%d" -d "last hour")
cur_hour=$(date -u +"%H" -d "last hour")

for fn in $(hdfs dfs -ls /mysql_audit/${cur_date}/*_${cur_hour}.gz|awk \'{print $NF}\');do
    host_name=$(echo $fn|awk -F [/_] \'{print $(NF-2)}\')
    module=${host_name%-bjc*}
    echo "load data inpath \'hdfs://ossmondb${fn}\' into table mysql_audit partition(dt=${cur_date_short},hour=${cur_hour},module=\'${module}\');" >> hive.sql
done

hive -f hive.sql && rm -f hive.sql

有任何疑问欢迎交流