Freeing Disk Space in Linux
Tags: amazon, cloud, Debian, disk, ec2, ext, ext2, ext3, ext4, filesystem, linux, nas, redhat, RHEL, san, space, storage, sysadmin, ubuntu, unix
Did you know that most filesystems reserve a percentage of the available free space as an emergency reserve for when the disk becomes full? This is a great safety mechanism if you’re running critical applications or database, but in many cases all this reserved space winds up going to waste. Especially so in the case of today’s 2 & 3 Terabyte disks!
On a linux EXT filesystem, 5% is reserved for access only by the root user. Assuming you have a 2T disk this is approximately 100G reservation, which is total overkill if you ask me! Luckily it’s easy enough to adjust on-the-fly with the tune2fs command.
For this example I’m going to change the number of reserved blocks from the default of 5% to 1% on the filesystem mounted as /misc.
A simple df -h
will show you free space minus the reserve.
[root@foo ~]# df -h | grep misc Filesystem Size Used Avail Use% Mounted on /dev/sda5 789G 110G 639G 15% /misc |
As you can see, there is 639G of free space on this filesystem.
Now, we use tune2fs -m 1
to change the percentage reserved to 1%
[root@foo ~]# tune2fs -m 1 /dev/sda5 tune2fs 1.39 (29-May-2006) Setting reserved blocks percentage to 1% (2133291 blocks) |
Now we check the available space
[root@foo ~]# df -h | grep misc Filesystem Size Used Avail Use% Mounted on /dev/sda5 789G 110G 672G 14% /misc |
Now we have 672G free. That’s 33G of storage we just got back for free!
Now, it may seem attractive to set this to 0% to squeeze the most possible free space from your drive, and under some circumstances this is perfectly safe to do. However, don’t set this to 0% if you have processes running as root that you want to continue running if the disk fills. Here’s a simple of thumb if you’re not sure. If you’ll loose irreplaceable data (or your job) if this filesystem goes casters up, don’t set to 0%.