linux集群相同目录文件差异比较脚本

#!/bin/bash

echo "######################################"
echo "检测多台服务器指定目录下的文件一致性"
echo "#####################################"

#通过对比两台服务器上文件的md5值，达到检测一致性的目的

if [ $# -lt 1 ]
then
    echo "请输入需要比较的文件目录或者文件"
    exit
fi

dir=$1

# 清空之前的文件夹
n_match=/tmp/xmatch/
rm -rf /tmp/xmatch/*
mkdir -p $n_match

# 遍历集群所有机器
hosts=(hadoop001 hadoop002 hadoop003 hadoop004)
for host in ${hosts[*]}
do
    echo ===================== $host ===================
    tmp_f="/tmp/md5_${host}.txt"
    ssh $host "if [ -f $tmp_f ]; then rm -f $tmp_f > /dev/null; fi"

    # 遍历所有目录，计算文件MD5
    for file in $@
    do
        # 判断文件是否存在
        if [ -e $file ];
        then
            # 获取父目录
            pdir=$(cd -P $(dirname $file); pwd)
            # 获取当前文件名
            fname=$(basename $file)
            full_file=$pdir/$fname

            ssh $host "if [ -f $full_file ];
            then
                echo $(md5sum $full_file) >> $tmp_f;
            else 
                find $pdir -type f|xargs md5sum >> $tmp_f;
            fi"
        else 
            echo ${file}文件不存在！
        fi
    done 
    # 将生成的文件复制到当前服务器
    scp $host:$tmp_f $n_match
    chmod 777 ${n_match}md5_${host}.txt
done

for((i=0;i<${#hosts[*]}-1;i++))
do
    for((j=1+$i;j<${#hosts[@]};j++))
    do
        echo ====================== 比较 ${hosts[i]} 和 ${hosts[j]} 的文件差异 =====================
        #diff ${n_match}md5_${hosts[i]}.txt ${n_match}md5_${hosts[j]}.txt -y -W 50
        for f in `awk '{print $2}' ${n_match}md5_${hosts[i]}.txt`
        do 
            if grep -qw "$f" ${n_match}md5_${hosts[j]}.txt;
            then 
                md5_a=`grep -w "$f" ${n_match}md5_${hosts[i]}.txt | awk '{print $1}'`
                md5_b=`grep -w "$f" ${n_match}md5_${hosts[j]}.txt | awk '{print $1}'`
                if [[ "$md5_a" != "$md5_b" ]];
                then
                    echo "${f} is changed！"
                fi
            else 
                echo "${f} is deleted."
            fi
        done
    done
done