Managing Basic Storage

Disk partitioning allows system administrators to divide a hard drive into multiple logical storage units, referred to as partitions. By separating a disk into partitions, system administrators can use different partitions to perform different functions.

For example, disk partitioning is necessary or beneficial in these situations:

  • Limit available space to applications or users.
  • Separate operating system and program files from user files.
  • Create a separate area for memory swapping.
  • Limit disk space use to improve the performance of diagnostic tools and backup imaging.

Since 1982, the Master Boot Record (MBR) partitioning scheme has dictated how disks are partitioned on systems running BIOS firmware. This scheme supports a maximum of four primary partitions. On Linux systems, with the use of extended and logical partitions, administrators can create a maximum of 15 partitions. Because partition size data is stored as 32-bit values, disks partitioned with the MBR scheme have a maximum disk and partition size of 2 TiB.

Because physical disks are getting larger, and SAN-based volumes even larger than that, the 2 TiB disk and partition size limit of the MBR partitioning scheme is no longer a theoretical limit, but rather a real-world problem that system administrators encounter more and more frequently in production environments. As a result, the legacy MBR scheme is in the process of being superseded by the new GUID Partition Table (GPT) for disk partitioning.

For systems running Unified Extensible Firmware Interface (UEFI) firmware, GPT is the standard for laying out partition tables on physical hard disks. GPT is part of the UEFI standard and addresses many of the limitations that the old MBR-based scheme imposes.

A GPT provides a maximum of 128 partitions. Unlike an MBR, which uses 32 bits for storing logical block addresses and size information, a GPT allocates 64 bits for logical block addresses. This allows a GPT to accommodate partitions and disks of up to eight zebibytes (ZiB) or eight billion tebibytes.

In addition to addressing the limitations of the MBR partitioning scheme, a GPT also offers some additional features and benefits. A GPT uses a globally unique identifier (GUID) to identify each disk and partition. In contrast to an MBR, which has a single point of failure, a GPT offers redundancy of its partition table information. The primary GPT resides at the head of the disk, while a backup copy, the secondary GPT, is housed at the end of the disk. A GPT uses a checksum to detect errors and corruptions in the GPT header and partition table.

Partition editors are programs which allow administrators to make changes to a disk's partitions, such as creating partitions, deleting partitions, and changing partition types. To perform these operations, administrators can use the Parted partition editor for both the MBR and the GPT partitioning scheme.

The parted command takes the device name of the whole disk as the first argument and one or more subcommands. The following example uses the print subcommand to display the partition table on the /dev/vda disk.

parted /dev/vda print
Model: Virtio Block Device (virtblk)
Disk /dev/vda: 53.7GB
Sector size (logical/physical): 512B/512B
Partition Table: msdos
Disk Flags:

Number  Start   End     Size    Type     File system  Flags
 1      1049kB  10.7GB  10.7GB  primary  xfs          boot
 2      10.7GB  53.7GB  42.9GB  primary  xfs

If you do not provide a subcommand, parted opens an interactive session for issuing commands.

parted /dev/vda
GNU Parted 3.2
Using /dev/vda
Welcome to GNU Parted! Type 'help' to view a list of commands.
(parted) print
Model: Virtio Block Device (virtblk)
Disk /dev/vda: 53.7GB
Sector size (logical/physical): 512B/512B
Partition Table: msdos
Disk Flags:

Number  Start   End     Size    Type     File system  Flags
 1      1049kB  10.7GB  10.7GB  primary  xfs          boot
 2      10.7GB  53.7GB  42.9GB  primary  xfs

(parted) quit

By default, parted displays all the sizes in powers of 10 (KB, MB, GB). You can change that default with the unit subcommand which accepts the following parameters:

  • s for sector
  • B for byte
  • MiB, GiB, or TiB (powers of 2)
  • MB, GB, or TB (powers of 10)

parted /dev/vda unit s print
Model: Virtio Block Device (virtblk)
Disk /dev/vda: 104857600s
Sector size (logical/physical): 512B/512B
Partition Table: msdos
Disk Flags:

Number  Start      End         Size       Type     File system  Flags
 1      2048s      20971486s   20969439s  primary  xfs          boot
 2      20971520s  104857535s  83886016s  primary  xfs

As shown in the example above, you can also specify multiple subcommands (here, unit and print) on the same line.

To partition a new drive, you first have to write a disk label to it. The disk label indicates which partitioning scheme to use.

NOTE: Keep in mind that parted makes the changes immediately. A mistake with parted could definitely lead to data loss.

As the root user, use the following command to write an MBR disk label to a disk.

parted /dev/vdb mklabel msdos

To write a GPT disk label, use the following command.

parted /dev/vdb mklabel gpt

WARNING: The mklabel subcommand wipes the existing partition table. Only use mklabel when the intent is to reuse the disk without regard to the existing data. If a new label changes the partition boundaries, all data in existing file systems will become inaccessible.

Creating an MBR disk partition involves several steps:

1. Specify the disk device to create the partition on.

As the root user, execute the parted command and specify the disk device name as an argument. This starts the parted command in interactive mode and displays a command prompt.

parted /dev/vdb
GNU Parted 3.2
Using /dev/vdb
Welcome to GNU Parted! Type 'help' to view a list of commands.
(parted) 

2. Use the mkpart subcommand to create a new primary or extended partition.

(parted) mkpart
Partition type?  primary/extended? primary

NOTE: For situations where you need more than four partitions on an MBR-partitioned disk, create three primary partitions and one extended partition. This extended partition serves as a container within which you can create multiple logical partitions.

3. Indicate the file-system type that you want to create on the partition, such as xfs or ext4. This does not create the file system on the partition; it is only an indication of the partition type.

File system type?  [ext2]? xfs

To get the list of the supported file-system types, use the following command:

parted /dev/vdb help mkpart
  mkpart PART-TYPE [FS-TYPE] START END     make a partition

    PART-TYPE is one of: primary, logical, extended
    FS-TYPE is one of: btrfs, nilfs2, ext4, ext3, ext2, fat32, fat16, hfsx,
    hfs+, hfs, jfs, swsusp, linux-swap(v1), linux-swap(v0), ntfs, reiserfs,
    hp-ufs, sun-ufs, xfs, apfs2, apfs1, asfs, amufs5, amufs4, amufs3,
    amufs2, amufs1, amufs0, amufs, affs7, affs6, affs5, affs4, affs3, affs2,
    affs1, affs0, linux-swap, linux-swap(new), linux-swap(old)
    START and END are disk locations, such as 4GB or 10%.  Negative values
    count from the end of the disk.  For example, -1s specifies exactly the
    last sector.

    'mkpart' makes a partition without creating a new file system on the
    partition.  FS-TYPE may be specified to set an appropriate partition
    ID.

4. Specify the sector on the disk that the new partition starts on.

Start? 2048s

Notice the s suffix to provide the value in sectors. You can also use the MiB, GiB, TiB, MB, GB, or TB suffixes. If you do not provide a suffix, MB is the default. parted may round the value you provide to satisfy disk constraints.

When parted starts, it retrieves the disk topology from the device. For example, the disk physical block size is typically a parameter that parted collects. With that information, parted ensures that the start position you provide correctly aligns the partition with the disk structure. Correct partition alignment is important for optimal performance. If the start position results in a misaligned partition, parted displays a warning. With most disks, a start sector that is a multiple of 2048 is a safe assumption.

5. Specify the disk sector where the new partition should end.

End? 1000MB

With parted, you cannot directly provide the size of your partition, but you can quickly compute it with the following formula:

Size = End - Start

As soon as you provide the end position, parted updates the partition table on the disk with the new partition details.

6. Exit parted.

(parted) quit
Information: You may need to update /etc/fstab.

7. Run the udevadm settle command. This command waits for the system to detect the new partition and to create the associated device file under the /dev directory. It only returns when it is done.

udevadm settle

As an alternative to the interactive mode, you can also create the partition as follows:

parted /dev/vdb mkpart primary xfs 2048s 1000MB

The GTP scheme also uses the parted command to create new partitions:

1. Specify the disk device to create the partition on.

As the root user, execute the parted command with the disk device as the only argument to start parted in interactive mode with a command prompt.

parted /dev/vdb
GNU Parted 3.2
Using /dev/vdb
Welcome to GNU Parted! Type 'help' to view a list of commands.
(parted) 

2. Use the mkpart subcommand to start creating the new partition.

With the GPT scheme, each partition is given a name.

(parted) mkpart
Partition name?  []? usersdata

3. Indicate the file system type that you want to create on the partition, such as xfs or ext4. This does not create the file system on the partition; it is only an indication of the partition type.

File system type?  [ext2]? xfs

4. Specify the sector on the disk that the new partition starts on.

Start? 2048s

5. Specify the disk sector where the new partition should end.

End? 1000MB

As soon as you provide the end position, parted updates the partition table on the disk with the new partition details.

6. Exit parted.

(parted) quit
Information: You may need to update /etc/fstab.

7. Run the udevadm settle command. This command waits for the system to detect the new partition and to create the associated device file under the /dev directory. It only returns when it is done.

udevadm settle

As an alternative to the interactive mode, you can also create the partition as follows:

parted /dev/vdb mkpart usersdata xfs 2048s 1000MB

The following steps apply for both the MBR and GPT partitioning schemes.

1. Specify the disk that contains the partition to be removed.

As the root user, execute the parted command with the disk device as the only argument to start parted in interactive mode with a command prompt.

parted /dev/vdb
GNU Parted 3.2
Using /dev/vdb
Welcome to GNU Parted! Type 'help' to view a list of commands.
(parted) 

2. Identify the partition number of the partition to delete.

(parted) print
Model: Virtio Block Device (virtblk)
Disk /dev/vdb: 5369MB
Sector size (logical/physical): 512B/512B
Partition Table: gpt
Disk Flags:

Number  Start   End     Size   File system  Name       Flags
 1      1049kB  1000MB  999MB  xfs          usersdata

3. Delete the partition.

(parted) rm 1

The rm subcommand immediately deletes the partition from the partition table on the disk.

4. Exit parted.

(parted) quit
Information: You may need to update /etc/fstab.

After the creation of a block device, the next step is to add a file system to it. Red Hat Enterprise Linux supports many different file system types, but two common ones are XFS and ext4. Anaconda, the installer for Red Hat Enterprise Linux, uses XFS by default.

As root, use the mkfs.xfs command to apply an XFS file system to a block device. For ext4, use mkfs.ext4.

mkfs.xfs /dev/vdb1
meta-data=/dev/vdb1              isize=512    agcount=4, agsize=60992 blks
         =                       sectsz=512   attr=2, projid32bit=1
         =                       crc=1        finobt=1, sparse=1, rmapbt=0
         =                       reflink=1
data     =                       bsize=4096   blocks=243968, imaxpct=25
         =                       sunit=0      swidth=0 blks
naming   =version 2              bsize=4096   ascii-ci=0, ftype=1
log      =internal log           bsize=4096   blocks=1566, version=2
         =                       sectsz=512   sunit=0 blks, lazy-count=1
realtime =none                   extsz=4096   blocks=0, rtextents=0

After you have added the file system, the last step is to mount the file system to a directory in the directory structure. When you mount a file system onto the directory hierarchy, user-space utilities can access or write files on the device.

Administrators use the mount command to manually attach the device onto a directory location, or mount point. The mount command expects the device, the mount point, and optionally file system options as arguments. The file-system options customize the behavior of the file system.

mount /dev/vdb1 /mnt

You also use the mount command to view currently mounted file systems, the mount points, and the options.

mount | grep vdb1
/dev/vdb1 on /mnt type xfs (rw,relatime,seclabel,attr2,inode64,noquota)

Manually mounting a file system is a good way to verify that a formatted device is accessible and working as expected. However, when the server reboots, the system does not automatically mount the file system onto the directory tree again; the data is intact on the file system, but users cannot access it.

To make sure that the system automatically mounts the file system at system boot, add an entry to the /etc/fstab file. This configuration file lists the file systems to mount at system boot.

/etc/fstab is a white-space-delimited file with six fields per line.

cat /etc/fstab

#
# /etc/fstab
# Created by anaconda on Wed Feb 13 16:39:59 2019
#
# Accessible filesystems, by reference, are maintained under '/dev/disk/'.
# See man pages fstab(5), findfs(8), mount(8) and/or blkid(8) for more info.
#
# After editing this file, run 'systemctl daemon-reload' to update systemd
# units generated from this file.
#
UUID=a8063676-44dd-409a-b584-68be2c9f5570   /        xfs   defaults   0 0
UUID=7a20315d-ed8b-4e75-a5b6-24ff9e1f9838   /dbdata  xfs   defaults   0 0

When you add or remove an entry in the /etc/fstab file, run the systemctl daemon-reload command, or reboot the server, for systemd to register the new configuration.

systemctl daemon-reload

The first field specifies the device. This example uses the UUID to specify the device. File systems create and store the UUID in their super block at creation time. Alternatively, you could use the device file, such as /dev/vdb1.

NOTE: Using the UUID is preferable because block device identifiers can change in certain scenarios, such as a cloud provider changing the underlying storage layer of a virtual machine, or the disks being detected in a different order with each system boot. The block device file name may change, but the UUID remains constant in the file system's super block.

Use the lsblk –fs command to scan the block devices connected to a machine and retrieve the file system UUIDs.

lsblk --fs
NAME   FSTYPE LABEL UUID                                 MOUNTPOINT
sr0
vda
└─vda1 xfs          a8063676-44dd-409a-b584-68be2c9f5570 /
vdb
└─vdb1 xfs          7a20315d-ed8b-4e75-a5b6-24ff9e1f9838 /dbdata

The second field is the directory mount point, from which the block device will be accessible in the directory structure. The mount point must exist; if not, create it with the mkdir command.

The third field contains the file-system type, such as xfs or ext4.

The fourth field is the comma-separated list of options to apply to the device. defaults is a set of commonly used options. The mount(8) man page documents the other available options.

The fifth field is used by the dump command to back up the device. Other backup applications do not usually use this field.

The last field, the fsck order field, determines if the fsck command should be run at system boot to verify that the file systems are clean. The value in this field indicates the order in which fsck should run. For XFS file systems, set this field to 0 because XFS does not use fsck to check its file-system status. For ext4 file systems, set it to 1 for the root file system and 2 for the other ext4 file systems. This way, fsck processes the root file system first and then checks file systems on separate disks concurrently, and file systems on the same disk in sequence.

NOTE: Having an incorrect entry in /etc/fstab may render the machine non-bootable. Administrators should verify that the entry is valid by unmounting the new file system and using mount /mountpoint, which reads /etc/fstab, to remount the file system. If the mount command returns an error, correct it before rebooting the machine.

As an alternative, you can use the findmnt –verify command to control the /etc/fstab file.

  1. parted(8)
  2. mkfs(8)
  3. mount(8)
  4. lsblk(8)
  5. fstab(5)