小知识:Linux如何删除大量小文件

环境:RHEL 6.5 + Oracle 11.2.0.4
需求:使用df -i巡检发现Inodes使用率过高,需要清理删除文件来解决。如果Inodes满,该目录将不能写,即使df -h查看还有剩余空间。

1.问题现象

Oracle的adump下记录的是sys的登陆审计信息,特点是小碎文件非常多,经常会遇到使用rm -rf *命令删除不了,报错-bash: /bin/rm: Argument list too long
这是因为通配符*在执行时会替换为具体的文件名,例如rm -rf file1 file2 file3 ...,如果文件数量过多,就容易出现这个错误。
比如在下面的环境中,adump目录下文件已达到114万+,执行rm -rf *命令时就会报这个错误:

[oracle@jystdrac2 adump]$ pwd
/opt/app/oracle/admin/crmdb/adump
[oracle@jystdrac2 adump]$ ls|wc -l
1149787
[oracle@jystdrac2 adump]$ rm -rf *
-bash: /bin/rm: Argument list too long
[oracle@jystdrac2 adump]$ du -sh
4.4G    

2.解决方案

清楚了问题现象,解决方案就从除去rm -rf *命令的方式之外,还有哪些方法可用,如果通过网络搜索,可能会找到结合find命令再去执行rm的方式,但其实效率非常差,具体写法这里就不列出了,因为我们通常也不会这样处理。那么如何较为效率的删除大批小文件呢?结合网络的经验,并实测验证,最终总结了两种常见的解决方案,效率上也都尚可。

方案一:巧用rsync的方式达到删除目的

建立一个空文件夹,使用rsync --delete-before -d <空文件夹> <需要清理删除小文件的目录>命令最终达到删除大批小文件的目的。下面演示具体操作:

[oracle@jystdrac2 adump]$ mkdir /data/null
[oracle@jystdrac2 adump]$ ls -l /data/null
total 0
[oracle@jystdrac2 ~]$ nohup rsync --delete-before -d /data/null/ /opt/app/oracle/admin/crmdb/adump/ &

使用man rsync查看rsync命令相关的参数说明如下:

-d, --dirs                  transfer directories without recursing
--delete-before         receiver deletes before transfer (default)

方案二:使用find命令的delete参数

使用find <需要清理删除小文件的目录> -type f -delete命令直接删除大批小文件。

使用man find查看find命令相关的参数说明如下:

       -type c
              File is of type c:

              b      block (buffered) special

              c      character (unbuffered) special

              d      directory

              p      named pipe (FIFO)

              f      regular file

              l      symbolic  link; this is never true if the -L option or the -follow option is in effect, unless the symbolic link is broken.  If you want to
                     search for symbolic links when -L is in effect, use -xtype.

              s      socket

              D      door (Solaris)

       -delete
              Delete files; true if removal succeeded.  If the removal failed, an error message is issued.  If -delete fails, find’s exit status will be nonzero
              (when it eventually exits).  Use of -delete automatically turns on the ‘-depth’ option.

              Warnings:  Don’t forget that the find command line is evaluated as an expression, so putting -delete first will make find try to delete everything
              below the starting points you specified.  When testing a find command line that you later intend to use with -delete, you should explicitly  spec-
              ify -depth in order to avoid later surprises.  Because -delete implies -depth, you cannot usefully use -prune and -delete together.

下面演示具体操作:

[oracle@jystdrac1 adump]$ nohup find /opt/app/oracle/admin/crmdb/adump/ -type f -delete &

可以参考下面的命令来简单监控删除过程中Inodes使用率的变化:

while true; do df -i /; sleep 10; done

比如我这里节点jystdrac1使用的find方法,节点jystdrac2使用的rsync方法,实际观察Inodes释放速度区别并不大:

# 使用的find方法,观察Inodes释放速度:
[oracle@jystdrac1 ~]$ while true; do df -i /; sleep 10; done
Filesystem                        Inodes   IUsed  IFree IUse% Mounted on
/dev/mapper/vg_linuxbase-lv_root 1806896 1519124 287772   85% /
Filesystem                        Inodes   IUsed  IFree IUse% Mounted on
/dev/mapper/vg_linuxbase-lv_root 1806896 1519015 287881   85% /
Filesystem                        Inodes   IUsed  IFree IUse% Mounted on
/dev/mapper/vg_linuxbase-lv_root 1806896 1513880 293016   84% /
Filesystem                        Inodes   IUsed  IFree IUse% Mounted on
/dev/mapper/vg_linuxbase-lv_root 1806896 1511132 295764   84% /
Filesystem                        Inodes   IUsed  IFree IUse% Mounted on
/dev/mapper/vg_linuxbase-lv_root 1806896 1502434 304462   84% /
Filesystem                        Inodes   IUsed  IFree IUse% Mounted on
/dev/mapper/vg_linuxbase-lv_root 1806896 1494583 312313   83% /
Filesystem                        Inodes   IUsed  IFree IUse% Mounted on
/dev/mapper/vg_linuxbase-lv_root 1806896 1489111 317785   83% /
Filesystem                        Inodes   IUsed  IFree IUse% Mounted on
/dev/mapper/vg_linuxbase-lv_root 1806896 1487629 319267   83% /

# 使用的rsync方法,观察Inodes释放速度:
[oracle@jystdrac2 ~]$ while true; do df -i /; sleep 10; done
Filesystem                        Inodes  IUsed  IFree IUse% Mounted on
/dev/mapper/vg_linuxbase-lv_root 1806896 963029 843867   54% /
Filesystem                        Inodes  IUsed  IFree IUse% Mounted on
/dev/mapper/vg_linuxbase-lv_root 1806896 955037 851859   53% /
Filesystem                        Inodes  IUsed  IFree IUse% Mounted on
/dev/mapper/vg_linuxbase-lv_root 1806896 953088 853808   53% /
Filesystem                        Inodes  IUsed  IFree IUse% Mounted on
/dev/mapper/vg_linuxbase-lv_root 1806896 950523 856373   53% /
Filesystem                        Inodes  IUsed  IFree IUse% Mounted on
/dev/mapper/vg_linuxbase-lv_root 1806896 948754 858142   53% /
Filesystem                        Inodes  IUsed  IFree IUse% Mounted on
/dev/mapper/vg_linuxbase-lv_root 1806896 944613 862283   53% /
Filesystem                        Inodes  IUsed  IFree IUse% Mounted on
/dev/mapper/vg_linuxbase-lv_root 1806896 942619 864277   53% /
Filesystem                        Inodes  IUsed  IFree IUse% Mounted on
/dev/mapper/vg_linuxbase-lv_root 1806896 938510 868386   52% /

既然两种方式差异不算大,那就根据需求或个人习惯选择即可。我自己更倾向于使用方案二,因为这样无需创建空目录,操作上也更直观。
最后再总结下删除大量小文件的方法:

# 方案一:
mkdir <空文件夹>
rsync --delete-before -d <空文件夹> <需要清理删除小文件的目录>
# 方案二:
find <需要清理删除小文件的目录> -type f -delete

相对来说这两种方式都比较效率,但由于整体小文件也是比较多,所以实际可以选择nohup放到后台执行。

This entry was posted in Linux and tagged , , , , . Bookmark the permalink.