Linux File Systems (mkfs, mount, fstab)

This article provides an introduction to Linux file systems, with specific reference to the information needed for the RHCSA EX200 and RHCE EX300 certification exams.

Remember, the exams are hands-on, so it doesn't matter which method you use to achieve the result, so long as the end product is correct.

mkfs

If you have used the "Disk Utility" to create your partitions, or they were created during the installation by "Disk Druid", they will already be formatted with a file system, so this step will be unnecessary.

Once you have created a suitable partition, it is likely you will want to create a file system on it. This can be done using one of the variations of the mkfs command. There are a wide variety of file system types available, but the Red Hat certification exams only consider ext2, ext3 and ext4. The ext4 file system is the default for RHEL6, so I will focus on that. The mkfs command destroys any data on the partition, so be very careful when issuing the command. The following three commands all format and ext4 file system on the specified partition.

# mkfs.ext4 /dev/sdb3
# mkfs -t ext4 /dev/sdb3
# mke2fs -t ext4 /dev/sdb3

Not surprisingly, to create an ext2 or ext3 file system simply replace ext4 in the above commands.

Although not required for the certification exam, the xfs file system is quite important for single instance Oracle installations. If you are storing datafiles on a cooked file system, as opposed to ASM, you should probably consider using xfs, rather than any of the ext* file systems.

The RHEL6 Storage Administration Guide contains more information about file system options for the ext* and xfs file systems, but these are not part of the exam syllabus.

mount and umount

Once a file system has been created, it can be mounted using the mount command.

mount -t type device dir
mount -t type -o options device dir

For example, the following output shows the creation of a new directory, which is used as a mount point for the "/dev/sdb3" ext4 volume.

# mkdir /u02
# mount -t ext4 -o rw /dev/sdb3 /u02
# df -h
Filesystem            Size  Used Avail Use% Mounted on
/dev/mapper/vg_rhce1-lv_root
                       26G  3.0G   21G  13% /
tmpfs                1002M  260K 1002M   1% /dev/shm
/dev/sda1             485M   52M  409M  12% /boot
/dev/sdb3             1.9G   35M  1.8G   2% /u02
#

The mount will only last until the next reboot, unless an entry is placed in the "/etc/fstab".

The umount command is used to unmount file systems.

# umount /u02
# df -h
Filesystem            Size  Used Avail Use% Mounted on
/dev/mapper/vg_rhce1-lv_root
                       26G  3.0G   21G  13% /
tmpfs                1002M  272K 1002M   1% /dev/shm
/dev/sda1             485M   52M  409M  12% /boot
#

/etc/fstab

Adding entries into the "/etc/fstab" allows you to make mounts persistent between reboots. If you have used the mount command, much of the entries will look familiar. The basic format of an entry is shown below.

fs_spec  fs_file  fs_vfstype  fs_mntops  fs_freq  fs_passno

These elements are explained below.

fs_spec : The device name, UUID or the label of the volume to be mounted. Remote file systems are referenced in a slightly different manner.
fs_spec : The mount point.
fs_vfstype : The type of file system (ext4, xfs, nfs, cifs etc.).
fs_mntops : Any required mount options.
fs_freq : When set to "1", data is automatically dumped by the dump command on shutdown. When set to "0", it is not.
fs_passno : Used by fsck to determine the order of file system checks on reboot. Set the root file system to "1" and other local file systems to "2". Remote or removable file systems should be set to "0".

Before looking at some sample entries, it's worth understanding how the UUID and volume labels can be identified. The blkid command displays the UUID of the specified device, or all devices if no device is specified.

# blkid /dev/sdb3
/dev/sdb3: UUID="e98bc512-f45c-4a93-ac84-d96b7d9f7750" TYPE="ext4"
#

If you have set a volume label, it will also be displayed.

# e2label /dev/sdb3 "Data_Volume"
# blkid /dev/sdb3
/dev/sdb3: UUID="e98bc512-f45c-4a93-ac84-d96b7d9f7750" TYPE="ext4" LABEL="Data_Volume" 
#

The following output shows typical fstab entries for a variety of mounts.

# Mount using device name
/dev/sdb3	                           /u02    ext4    defaults 1 2

# Mount using UUID
UUID=e98bc512-f45c-4a93-ac84-d96b7d9f7750  /u02    ext4    defaults 1 2

# Mount using label
LABEL=Data_Volume                          /u02    ext4    defaults 1 2

# Mount an NFS share.
123.123.123.123:/nfs_share                 /u02    nfs     defaults 0 0

# Mount a samba (CIFS) share using in-line credentials or a credentials file.
//123.123.123.123/share	                   /u02    cifs    rw,username=my_user,password=mypassword 0 0
//123.123.123.123/share	                   /u02    cifs    rw,credentials=/root/.smbcred 0 0

# Mount an ISO file.
/u01/myfile.iso                            /u02    df,iso9660 loop,ro,auto 0 0

# Setting the ownership of the mount point
//123.123.123.123/share                    /u02    cifs    rw,credentials=/root/.smbcred,uid=500,guid=500 0 0

For more information about using a credentials file, look here.

The are a variety of mount options and the ones you choose will depend on what you are trying to achieve. Vendors like Oracle will often specify required options for specific file systems.

Swap

Swap partitions are managed a little differently than regular file systems. A swap partition is formatted using the mkswap command.

# mkswap /dev/sdb2
Setting up swapspace version 1, size = 1951892 KiB
no label, UUID=83aa1e29-10c1-4821-9491-371df4edc5cb
#

Including the "-L" option will allow you to label the swap partition at the same time.

# mkswap -L swap_vol /dev/sdb2
Setting up swapspace version 1, size = 1951892 KiB
LABEL=swap_vol, UUID=eee7cd19-fe63-4bf9-a085-835112099c8f
#

The swap partition is then enabled using the swapon command. The "-v" flag is not necessary, but verbose mode is a little more reassuring as without it the command produces no output.

# swapon -v /dev/sdb2
swapon on /dev/sdb2
swapon: /dev/sdb2: found swap signature: version 1, page-size 4, same byte order
swapon: /dev/sdb2: pagesize=4096, swapsize=1998741504, devsize=1998743040
#

We can see the current swap volumes using the "-s" flag.

# swapon -s
Filename				Type		Size	Used	Priority
/dev/dm-1                               partition	4128760	0	-1
/dev/sdb2                               partition	1951888	0	-2
#

The swapoff command is used to disable a swap volume.

# swapoff /dev/sdb2
# swapon -s 
Filename				Type		Size	Used	Priority
/dev/dm-1                               partition	4128760	0	-1
#

Upgrading File Systems

The preferred option for any file system upgrade is to create a new file system and migrate the data to it, but this is not always possible. Before doing any file system upgrade, you should do the following things:

Unmount the file system.
Run the fsck command to make sure the file system is OK.
Perform a backup of the data on the file system.

To upgrade an existing ext2 file system to ext3, use the tune2fs command with the "-j" flag to turn on journalling.

# tune2fs -j /dev/sdb4
Creating journal inode: done
This filesystem will be automatically checked every 23 mounts or
180 days, whichever comes first.  Use tune2fs -c or -i to override.
#

To convert an existing ext3 file system to ext4 issue the tune2fs command with the following options.

# tune2fs -O extents,uninit_bg,dir_index /dev/sdb4
tune2fs -O extents,uninit_bg,dir_index /dev/sdb4
tune2fs 1.41.12 (17-May-2010)

Please run e2fsck on the filesystem.

#

In all attempted ext3 to ext4 conversions on OL6.2 I was prompted to check the file system, as seen above. The file system check repaired the issues in each case.

# e2fsck /dev/sdb4
e2fsck 1.41.12 (17-May-2010)
One or more block group descriptor checksums are invalid.  Fix<y>? yes

Group descriptor 0 checksum is invalid.  FIXED.
Group descriptor 1 checksum is invalid.  FIXED.
Group descriptor 2 checksum is invalid.  FIXED.
Group descriptor 3 checksum is invalid.  FIXED.
Group descriptor 4 checksum is invalid.  FIXED.
Group descriptor 5 checksum is invalid.  FIXED.
Group descriptor 6 checksum is invalid.  FIXED.
Group descriptor 7 checksum is invalid.  FIXED.
Group descriptor 8 checksum is invalid.  FIXED.
Group descriptor 9 checksum is invalid.  FIXED.
Group descriptor 10 checksum is invalid.  FIXED.
Group descriptor 11 checksum is invalid.  FIXED.
Group descriptor 12 checksum is invalid.  FIXED.
Group descriptor 13 checksum is invalid.  FIXED.
Group descriptor 14 checksum is invalid.  FIXED.
old_data contains a file system with errors, check forced.
Pass 1: Checking inodes, blocks, and sizes
Pass 2: Checking directory structure
Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Pass 5: Checking group summary information
old_data: 11/122160 files (0.0% non-contiguous), 16598/487974 blocks
#

After performing a file system upgrade, you should always do the following things.

Perform a file system check (fsck or e2fsck).
If the file system is present in the "/etc/fstab", amend the fs_vfstype entry so it is mounted as the correct file system type.

Checking File Systems

The fsck (or e2fsck) command is used to check an unmounted file system. Simply pass the device as a parameter.

# fsck /dev/sdb4
e2fsck 1.41.12 (17-May-2010)
old_vol: clean, 11/122160 files, 8397/487974 blocks
# 
# e2fsck /dev/sdb4
fsck from util-linux-ng 2.17.2
e2fsck 1.41.12 (17-May-2010)
old_vol: clean, 11/122160 files, 8397/487974 blocks
#

Remember, if the file system is mounted, it should be unmounted first (or mounted as read only). In the following example, the file system is mounted to the "/u02" mount point.

# umount /u02
# e2fsck /dev/sdb4
e2fsck 1.41.12 (17-May-2010)
old_vol: clean, 11/122160 files, 8397/487974 blocks
# mount /u02

Linux File Systems (mkfs, mount, fstab) ext4