What Is Eating CentOS Disk Space

scottalanmiller

Let's start by determining which filesystem is the culprit. We can track this down with:

df -h

Let us know the output of that command.

scottalanmiller

Once we know where the issue is, we can use the "du" command to dig into it. If we determine that the space is being used up on / (the root partition) then we can dig into that with this:

*du -shx / **

scottalanmiller

But if that returns a lot of data, that can be annoying, so a way to make that show up in order is like this:

du -smx * | sort -n

This will sort the output putting the biggest space users at the bottom of the list (so that the stuff that scrolls off the top is the little stuff that you don't care about.

scottalanmiller

So the process to track down the biggest problems are to start with df -h to determine which filesystem is the problem. Then start at the root of that filesystem and use du -smx * | sort -n to find the biggest space using directories there. Then cd into the directories and run du -smx * | sort -n again and keep looping through it like this until you find where space is being used that should not be.

ajin.c

df -h

Filesystem            Size  Used Avail Use% Mounted on
/dev/mapper/vg_trvbackup-lv_root
                       50G   48G     0 100% /
tmpfs                 3.9G     0  3.9G   0% /dev/shm
/dev/sda1             485M   53M  407M  12% /boot
/dev/mapper/vg_trvbackup-lv_home
                      402G  145G  236G  39% /home
/usr/tmpDSK           1.6G   37M  1.5G   3% /tmp
/dev/sdb1             1.5T  286G  1.2T  20% /backup/current
/dev/sdb2             322G  211G   96G  69% /backup/archive

ajin.c

du -shx /*

out put keeps on counting .........

36K     /backup
6.4M    /bin
43M     /boot
772K    /dev
29M     /etc

and so onn

ajin.c

root@trvbackup [~]# du -smx * | sort -n
1 anaconda-ks.cfg
1 CHANGELOG
1 cpanel3-skel
1 installer.lock
1 install.log
1 install.log.syslog
1 install.sh
1 latest
1 LICENSE
1 php.ini.new
1 php.ini.orig
1 public_ftp
1 public_html
1 README
1 scripts
1 tmp
3 csf

ajin.c

trying on it......

scottalanmiller

@ajin.c said:

du -shx /*

out put keeps on counting .........

36K /backup
6.4M /bin
43M /boot
772K /dev
29M /etc

and so onn

It takes a while if the system is full. The "and so on" is the part that is important.

scottalanmiller

@ajin.c said:

root@trvbackup [~]# du -smx * | sort -n
1 anaconda-ks.cfg
1 CHANGELOG
1 cpanel3-skel
1 installer.lock
1 install.log
1 install.log.syslog
1 install.sh
1 latest
1 LICENSE
1 php.ini.new
1 php.ini.orig
1 public_ftp
1 public_html
1 README
1 scripts
1 tmp
3 csf

You switched into root's home director "/root" which is not using any space. So this output won't help. You need to start at /. So do this...

cd /
du -smx * | sort -n

And provide the complete results.

scottalanmiller

Adding keywords for anyone searching later: CentOS RHEL Red Hat Enterprise Linux

scottalanmiller

Here is some sample output from a web server I happen to be logged into at the moment. I added the "2> /dev/null" and the "tail" portions to make it easier to read and use. Make sure you are root before doing this to make things easy.

[root@to-lnx-web /]# **whoami**
root
[root@to-lnx-web /]# **pwd**
/
[root@to-lnx-web /]# **du -smx * 2> /dev/null| sort -n | tail -n 5**
153     boot
403     tmp
554     lib
899     usr
6070    var
[root@to-lnx-web /]# **cd /var**
[root@to-lnx-web var]# **du -smx * 2> /dev/null| sort -n | tail -n 5**
70      tmp
73      spool
184     lib
1708    www
3957    log
[root@to-lnx-web var]# **cd log**
[root@to-lnx-web log]# **du -smx * 2> /dev/null| sort -n | tail -n 5**
316     httpd
413     maillog-20140223
627     maillog
1043    maillog-20140302
1267    maillog-20140309

scottalanmiller

From my output above, you can see that I started in / and found that var was the directory using the most space under it. So I moved into var and did it again. Under var we saw that log was using the most space. So we moved until log and ran it again.

The 2>/dev/null removes extraneous error output that you don't care about.

The sort -n | tail -n 5 portion shows you only the five largest files or directories from each run. You could adult the "5" to "8" or "12" or whatever is most useful to you.

ajin.c

root@trvbackup [/]# du -smx * | sort -n
^C
root@trvbackup [/]#

Waited arround half an hour ...but no output ....still waiting

scottalanmiller

If the drive is full, this will likely take some time. Because it is sorting the output it will show nothing until it completes.

ajin.c

Boss.....Still waiting for the output.......

ajin.c

root@trvbackup [/]# du -smx * | sort -n
du: cannot access proc/11877/task/11877/fd/4': No such file or directory du: cannot access proc/11877/task/11877/fdinfo/4': No such file or directory
du: cannot access proc/11877/fd/4': No such file or directory du: cannot access proc/11877/fdinfo/4': No such file or directory
0 proc
0 scripts
0 sys
1 backup
1 dev
1 lost+found
1 media
1 mnt
1 quota.user
1 razor-agent.log
1 selinux
1 srv
3 tmp
7 bin
8 root
14 sbin
29 etc
30 lib64
38 opt
43 boot
234 lib
5401 usr
17480 var
148041 home

scottalanmiller

This is easy. It's someone storing stuff in their home directory. This is not a system problem but a user problem. Just just the same command but with /home instead of just / and it will produce the list of your offending users.

scottalanmiller

That is 148GB of user data.

ajin.c

root@trvbackup [/home]# du -smx * | sort -n

right ?