Linux服务器数据定期同步和备份方式

原创 2017-07-11 陈同生信宝典

数据安全是做数据分析的人需要关注的一大问题。对于我们分析的关键数据、使用的关键脚本都需要定期备份。

scp

最简单的备份方式，就是使用cp (本地硬盘)或scp (远程硬盘)命令，给自己的结果文件新建一个拷贝；每有更新，再拷贝一份。具体命令如下：

cp -fur source_project project_bak
scp -r source_project user@remote_server_ip:project_bak

为了实现定期备份，我们可以把上述命令写入crontab程序中，设置每天的晚上23:00执行。对于远程服务器的备份，我们可以配置免密码登录，便于自动备份。后台输入免密码登录服务器，获取免密码登录服务器的方法。

# Crontab format
# Minute　　Hour　　Day　　Month　　Week　　command 
# * 表示每分/时/天/月/周
# 每天23:00 执行cp命令
0          23      *       *       *      cp -fur source_project project_bak
# */2 表示每隔2分分/时/天/月/周执行命令
# 每隔24小时执行cp命令
0          */24      *       *       *      cp -fur source_project project_bak
0          0          */1     *        *     scp -r source_project user@remote_server_ip:project_bak

# 另外crotab还有个特殊的时间
# @reboot: 开机运行指定命令
@reboot cmd

rsync

cp或scp使用简单，但每次执行都会对所有文件进行拷贝，耗时耗力，尤其是需要拷贝的内容很多时，重复拷贝对时间和硬盘都是个损耗。

rsync则是一个增量备份工具，只针对修改过的文件的修改过的部分进行同步备份，大大缩短了传输的文件的数量和传输时间。具体使用如下：

# 把本地project目录下的东西备份到远程服务器的/backup/project目录下
# 注意第一个project后面的反斜线，表示拷贝目录内的内容，不在目标目录新建project文件夹。注意与第二个命令的比较，两者实现同样的功能。
# -a: archive mode, quals -rlptgoD
# -r: 递归同步
# -p: 同步时保留原文件的权限设置
# -u: 若文件在远端做过更新，则不同步，避免覆盖远端的修改
# -L: 同步符号链接链接的文件，防止在远程服务器出现文件路径等不匹配导致的软连接失效
# -t: 保留修改时间
# -v: 显示更新信息
# -z: 传输过程中压缩文件，对于传输速度慢时适用
rsync -aruLptvz --delete project/ user@remoteServer:/backup/project
rsync -aruLptvz --delete project user@remoteServer:/backup/

rsync所做的工作为镜像，保证远端服务器与本地文件的统一。如果本地文件没问题，远端也不会有问题。但如果发生误删或因程序运行错误，导致文件出问题，而在同步之前又没有意识到的话，远端的备份也就没了备份的意义，因为它也被损坏了。误删是比较容易发现的，可以及时矫正。但程序运行出问题，则不一定了。

rdiff-backup

这里推荐一个工具rdiff-backup不只可以做增量备份，而且会保留每次备份的状态，新备份和上一次备份的差别，可以轻松回到之前的某个版本。唯一的要求就是，本地服务器和远端服务器需要安装统一版本的rdiff-backup。另外还有2款工具 duplicity和`Rsnapshot也可以做类似工作，但方法不一样，占用的磁盘空间也不一样，具体可查看原文链接中的比较。

具体的rdiff-backup安装和使用如下 (之前写的是英文，内容比较简单，就不再翻译了)：

Install `rdiff-backup` at both local and remote computers

Install requirements
#install for ubuntu, debian sudo apt-get install python-dev librsync-dev #self compile #downlaod rsync-dev from https://sourceforge.net/project/showfiles.php?group_id=56125 tar xvzf librsync-0.9.7.tar.gz export CFLAGS="$CFLAGS -fPIC" ./configure --prefix=/home/user/rsync --with-pic make make install
Install rdiff-backup
#See Reference part for download link # http://www.nongnu.org/rdiff-backup/ python setup.py install --prefix=/home/user/rdiff-backup #If you complied rsync-dev yourself, please specify the location of rsync-dev python setup.py --librsync-dir=/home/user/rsync install -- prefix=/home/user/rdiff-backup
Add exeutable files and python modules to environmental variables
#Add the following words into .bashrc or .bash_profile or any other config files export PATH=${PATH}:/home/user/rdiff-backup/bin export PYTHONPATH=${PYTHONPATH}:/home/user/rdiff-backup/lib/python2.x/site-packages #pay attention to the x in python2.x of above line which can be 6 or 7 depending on #the Python version used.
Test environmental variable when executing commands through ssh
ssh user@host 'echo ${PATH}' #When I run this command in my local computer, #I found only system environmetal variable is used #and none of my self-defined environmetal variable is used. #Then, I modified the following lines in file 'SetConnections.py' in #/home/user/rdiff-backup/lib/python2.x/site-packages/rdiff_backup #to set environmental explicitly when login. #pay attention to the single quote used inside double quote __cmd_schema = "ssh -C %s 'source ~/.bash_profile; rdiff-backup --server'" __cmd_schema_no_compress = "ssh %s 'source ~/.bash_profile; rdiff-backup --server'" #choose the one contains environmental variable for rdiff-backup from .bash_profile and .bashrc.

Use `rdiff-backup`

Start backup

rdiff-backup --no-compression --print-statistics user@host::/home/user/source_dir destination_dir
If the destination_dir exists, please add --force like rdiff-backup --no-compression --force --print-statistics user@host::/home/user/source_dir destination_dir. All things in original destination_dir will be depleted.
If you want to exclude or include special files or dirs please specify like --exclude '**trash' or --include /home/user/source_dir/important.

Timely backup your data

Add the above command into crontab (hit 'crontab -e' in terminal to open crontab) in the format like 5 22 */1 * * command which means executing the command at 22:05 everyday.

Restore data

Restore the latest data by running rdiff-backup -r now destination_dir user@host::/home/user/source_dir.restore. Add --force if you want to restore to source_dir.
Restore files 10 days ago by running rdiff-backup -r 10D destination_dir user@host::/home/user/source_dir.restore. Other acceptable time formats include 5m4s (5 minutes 4 seconds) and 2014-01-01 (January 1st, 2014).
Restore files from an increment file by running rdiff-backup destination_dir/rdiff-backup-data/increments/server_add.2014-02-21T09:22:45+08:00.missing user@host::/home/user/source_dir.restore/server_add. Increment files are stored in destination_dir/rdiff-backup-data/increments/server_add.2014-02-21T09:22:45+08:00.missing.

Remove older records to save space

Deletes all information concerning file versions which have not been current for 2 weeks by running rdiff-backup --remove-older-than 2W --force destination_dir. Note that an existing file which has not changed for a year will still be preserved. But a file which was deleted 15 days ago can not be restored after this command. Normally one should use --force since it is used to delete multiple increments at the same time which --remove-older-than refuses to do by default.
Only keeps the last n rdiff-backup sessions by running rdiff-backup --remove-older-than 20B --force destination_dir.

Statistics

Lists increments in given golder by rdiff-backup --list-increments destination_dir/.
Lists of files changed in last 5 days by rdiff-backup --list-changed-since 5D destination_dir/.
Compare the difference between source and bak by rdiff-backup --compare user@host::source-dir destination_dir
Compare the sifference between source and bak (as it was two weeks ago) by rdiff-backup --compare-at-time 2W user@host::source-dir destination_dir.

A complete script (automatically sync using `crontab`)

#!/bin/bash

export PYTHONPATH=${PYTHONPATH}:/soft/rdiff_backup/lib/python2.7/site-packages/

rdiff-backup --no-compression -v5 --exclude '**trash' user@server::source/ bak_dir/

ret=$?
if test $ret -ne 0; then
    echo "Wrong in bak" | mutt -s "Wrong in bak" bak@mail.com
else
    echo "Right in bak" | mutt -s "Right in bak" bak@mail.com
fi

echo "Finish rdiff-backup $0 ---`date`---"  >>bak.log 2>&1

echo "`rdiff-backup --exclude '**trash' --compare-at-time 1D user@server::source/ bak_dir/`" | mutt -s "Lists of baked files" bak@mail.com

揭“盗卖尸体案”易律师凌晨发文：公安下一步就是找我，律师职业也许画上......

台湾值不值得美国出兵？美重量级人物和智库警告：美国无法承受失去台湾

倾家荡产求热度，实则只会步入更深贫困陷阱

当一座城市决定“砸锅卖铁”

初中生家长注意：今年将有初中生受益！