由于工作原因,要管理一些图片服务器,其中有几台是作为图片存放的机器,每台机器配有10T的硬盘,最近硬盘告急,都只剩下几百个G,所以要删除一些图片。
目前服务器有一亿多个文件,删过这种大量的小碎文件的朋友一定知道使用rm -rf的下场,效果非常的差,基本上看不到什么效果。通过find命令带delete的效果也不太好,后来使用了rsync将空目录同步过去的方式,效果比较好,一天大概能清理出400G的空间。
为了保证并发的效果,写了脚本充当守护进程
- #!/bin/sh
- #./delete.sh get_dir_path_cmd rsync_delete_cmd_process_total_count
- #./delete.sh "find /data1/sinawapcms/htdocs/p_w_picpaths/iask/2012/08/* -maxdepth 0" 20
- rsync_path=rsync
- empty_dir=/tmp/empty/
- get_rsync_process_num()
- {
- echo $(ps -ef | grep rsync | grep delete | wc -l);
- }
- #$1 is rsync cmd , $2 is rsync total process num
- do_rsync_cmd()
- {
- while test 1 = 1
- do
- if test $(get_rsync_process_num) -lt $2
- then
- ($1 >/dev/null 2>&1 &);
- break;
- else
- sleep 100;
- fi
- done
- }
- if ! ( $rsync_path --version > /dev/null 2>&1 )
- then
- echo rsync path error;
- exit 1;
- fi
- if ! ls -l $empty_dir > /dev/null 2>&1
- then
- if ! mkdir -p $empty_dir >/dev/null 2>&1
- then
- echo Permission Denied, make sure you have permission to mkdir -p /tmp/empty ;
- exit 3;
- fi
- fi
- if ! test $( ls -la $empty_dir | wc -l) = 3
- then
- echo Directoty \"$empty_dir\" is not empty, empty directory? ;
- printf "Enter (Y/N):";
- read action < /dev/tty;
- if test $action = "Y" || test $action = 'y'
- then
- if ! rm -rf ${empth_dir}/* >/dev/null 2>&1
- then
- echo Permission Denied, make sure you have permission to rm -rf $empty_dir;
- exit 3;
- fi
- else
- echo error, Directoty \"$empty_dir\" is not empty;
- exit 2;
- fi
- fi
- dir=$($1);
- #rsync --delete-before -a -H /tmp/empty/ /data1/sinawapcms/htdocs/p_w_picpaths/iask/2012/08/ff/ >/dev/null &
- for i in $dir
- do
- sleep 5;
- cmd="$rsync_path --delete-before -a -H $empty_dir ${i}/" ;
- (do_rsync_cmd "$cmd" $2);
- done
调用方式:
./delete.sh "find /data1/sinawapcms/htdocs/p_w_picpaths/iask/2012/08/* -maxdepth 0" 20 第一个参数用来获取我们要清空的目录 第二个参数限定并发数,防止服务器负载过高