LVM Architecture

In the LVM architecture, several layers can be distinguished. On the lowest layer, the storage devices are used. These can be any storage devices, such as complete disks, partitions, logical units (LUNs) on a storage-area network (SAN), and whatever else is made possible in modern storage topologies. In this chapter you learn how to use partitions as physical volumes, which is recommended practice. By using partitions instead of complete disk devices, it is easy for other tools to recognize that some storage has already been configured on the block device, which makes it less likely that misconfigurations are going to occur.

The storage devices need to be flagged as physical volumes, which makes them usable in an LVM environment and makes them usable by other utilities trying to gain access to the logical volume. A storage device that is a physical volume can be added to the volume group, which is the abstraction of all available storage. The “abstraction” means that the volume group is not something that is fixed, but it can be resized when needed, which makes it possible to add more space on the volume group level when volumes are running out of disk space. The idea is simple: If you are running out of disk space on a logical volume, you take available disk space from the volume group. And if there is no available disk space in the volume group, you just add it by adding a physical volume.

On top of the volume group are the logical volumes. Logical volumes do not act on disks directly but get their disk space from available disk space in the volume group. That means that a logical volume may consist of available storage from multiple physical volumes, which adds an important layer of additional flexibility to the storage configuration.

The actual file systems are created on the logical volumes. Because the logical volumes are flexible with regard to size, that makes the file systems flexible as well. If a file system is running out of disk space, it is relatively easy to extend the file system or to reduce it if the file system allows that. Note that in order to resize file systems when logical volumes are resized, the file systems must offer support for that.

Figure 15-1 gives an overview of the LVM architecture.

LVM Features

There are several reasons why LVM is great. The most important reason is that LVM offers a flexible solution for managing storage. Volumes are no longer bound to the restrictions of physical hard drives. If additional storage space is needed, the volume group can easily be extended by adding a new physical volume, so that disk space can be added to the logical volumes. It is also possible to reduce the size of a logical volume, but only if the file system that was created on that volume supports the feature of reducing the size of the file system. Ext4 supports growing and shrinking; XFS size can only be increased.

Another important reason why administrators like using LVM is the support for snapshots. A snapshot keeps the current state of a logical volume and can be used to revert to a previous situation or to make a backup of the file system on the logical volume if the volume is open. Using snapshots is essential in backup strategies.

LVM snapshots are created by copying the logical volume administrative data (the metadata) that describes the current state of files to a snapshot volume. As long as nothing changes, from the LVM snapshot metadata the original blocks in the original volume are addressed. When blocks are modified, the blocks containing the previous state of the file are copied over to the snapshot volume, which for that reason will grow. Using this method ensures that, by accessing an LVM snapshot volume, the exact state of the files as they were when the snapshot was created can be accessed. Because the snapshot will grow when files on the original volume change, when planning for snapshots, you should make sure that a sufficient amount of disk space is available. Also, snapshots are supposed to be temporary: once a snapshot has served its purpose, it can be removed.

A third important advantage of using LVM logical volumes is the option to replace failing hardware easily. If a hard disk is failing, data can be moved within the volume group (through the pvmove command), the failing disk can then be removed from the volume group, and a new hard disk can be added dynamically, without requiring any downtime for the logical volume itself.

Creating the Physical Volumes

Before you can use the LVM tools to create physical volumes (PVs), you should create a partition marked as the LVM partition type. This is basically the same procedure as described in Chapter 14, with the only difference being that before writing changes to disk, you need to change the partition type.

In fdisk and gdisk, you can use t from the menu to change the type. If you are using an MBR disk, the partition type is 8e. If you are using a GUID disk, use the partition type 8e00. In fdisk you can also just type lvm as the partition type to use, as the RHEL 9 version of fdisk supports using aliases. If you are using parted to create partitions, you need to use the set n lvm on command from within the parted interface (where n is the number of the partition you want to mark for use with LVM).

After creating the partition and flagging it as an LVM partition type, you need to use pvcreate to mark it as a physical volume. This writes some metadata to the partition, which allows it to be used in a volume group. The entire procedure is summarized in Exercise 15-1, in which you create a physical volume. Also see Example 15-2 for an overview of this procedure.

Example 15-2 Creating an LVM Partition in fdisk

[root@
[root@server1 ~]# fdisk /dev/sdd

Welcome to fdisk (util-linux 2.37.4).
Changes will remain in memory only, until you decide to write them.
Be careful before using the write command.


Command (m for help): g
Created a new GPT disklabel (GUID:
  3BCE8E49-EFDF-9144-ACD5-290F4FCCDA07).

Command (m for help): n
Partition number (1-128, default 1):
First sector (2048-41943006, default 2048):
Last sector, +/-sectors or +/-size{K,M,G,T,P} (2048-41943006, default
  41943006): +1G

Created a new partition 1 of type 'Linux filesystem' and of size 1
GiB.

Command (m for help): t
Selected partition 1
Partition type or alias (type L to list all): lvm
Changed type of partition 'Linux filesystem' to 'Linux LVM'.

Command (m for help): p
Disk /dev/sdd: 20 GiB, 21474836480 bytes, 41943040 sectors
Disk model: VMware Virtual S
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disklabel type: gpt
Disk identifier: 3BCE8E49-EFDF-9144-ACD5-290F4FCCDA07

Device     Start      End Sectors Size Type
/dev/sdd1   2048 2099199 2097152   1G Linux LVM

Command (m for help): w
The partition table has been altered.
Calling ioctl() to re-read partition table.
Syncing disks.

Example 15-3 Verifying the Physical Volume

[root@server1 ~]# pvcreate /dev/sdd1
  Physical volume "/dev/sdd1" successfully created.
[root@server1 ~]# pvs
  PV         VG   Fmt  Attr PSize   PFree
  /dev/sda2  rhel lvm2 a--  <19.00g    0
  /dev/sdd1       lvm2 ---    1.00g 1.00g

As an alternative to the pvs command, which shows a summary of the physical volumes and their attributes, you can use the pvdisplay command to show more details. Example 15-4 shows an example of the output of this command.

Example 15-4 Sample pvdisplay Command Output

[root@server1 ~]# pvdisplay /dev/sdd1
  "/dev/sdd1" is a new physical volume of "1.00 GiB"
  --- NEW Physical volume ---
  PV Name                /dev/sdd1
  VG Name
  PV Size                1.00 GiB
  Allocatable           NO
  PE Size               0
  Total PE              0
  Free PE               0
  Allocated PE          0
  PV UUID               cjdhpJ-bRh9-fg3B-KlPh-XQzD-unXV-ycVK36

If you want a compact overview of the current storage configuration on your server, you might also like the lsblk command. As shown in Example 15-5, this command gives a hierarchical overview of which disks and partitions are used in what LVM volume groups and logical volumes.

Example 15-5 Using lsblk for a Compact Overview of the Current Storage Configuration

[root@server1 ~]# lsblk
NAME          MAJ:MIN RM SIZE RO TYPE MOUNTPOINTS
sda             8:0    0  20G  0 disk
├─sda1          8:1    0   1G  0 part /boot
└─sda2          8:2    0  19G  0 part
  ├─rhel-root 253:0    0  17G  0 lvm  /
  └─rhel-swap 253:1    0   2G  0 lvm  [SWAP]
sdb             8:16   0  20G  0 disk
├─sdb1          8:17   0   1G  0 part
├─sdb2          8:18   0   1K  0 part
├─sdb5          8:21   0   1G  0 part
└─sdb6          8:22   0   1G  0 part
sdc             8:32   0  20G  0 disk
└─sdc1          8:33   0   1G  0 part /exercise
sdd             8:48   0  20G  0 disk
└─sdd1          8:49   0   1G  0 part
sr0            11:0    1   8G  0 rom  /repo
nvme0n1       259:0    0  20G  0 disk

Creating the Volume Groups

Now that the physical volume has been created, you can assign it to a volume group (VG). It is possible to add a physical volume to an existing volume group (which is discussed later in this chapter), but here you learn how to create a new volume group and add the physical volume to it. This is a simple one-command procedure. Just type vgcreate followed by the name of the volume group you want to create and the name of the physical device you want to add to it. So, if the physical volume name is /dev/sdd1, the complete command is vgcreate vgdata /dev/sdd1. You are completely free in your choice of name for the volume group. I like to start all volume group names with vg, which makes it easy to find the volume groups if there are many, but you are free to choose anything you like.

Between the previous section and the preceding paragraph, you have learned how to create a volume group in a two-step procedure where you first create the physical volume with the pvcreate command and then add the volume group using the vgcreate command. You can do this in a one-step procedure as well (where using a separate pvcreate command will not be necessary).

The one-step procedure is particularly useful for adding a complete disk device). If you want to add the disk /dev/sdc, for instance, just type vgcreate vgdata /dev/sdc to create a volume group vgdata that contains the /dev/sdc device. When you are doing this to add a device that has not been marked as a physical volume yet, the vgcreate utility will automatically flag it as a physical volume so that you can see it while using the pvs command.

When you’re creating volume groups, a physical extent size is used. The physical extent size defines the size of the building blocks used to create logical volumes. A logical volume always has a size that is a multiple of the physical extent size. If you need to create huge logical volumes, it is more efficient to use a big physical extent size. If you do not specify anything, a default extent size of 4 MiB is used. The physical extent size is always specified as a multiple of 2 MiB, with a maximum size of 128 MiB. Use the vgcreate -s option to specify the physical extent size you want to use.

After creating the volume group, you can request details about the volume group using the vgs command for a short summary, or the vgdisplay command to get more information. Example 15-6 shows an example of the output of the vgdisplay command.

Example 15-6 Showing Current Volume Group Properties

[root@server1 ~]# vgdisplay vgdata
  --- Volume group ---
  VG Name                vgdata
  System ID
  Format                 lvm2
  Metadata Areas        1
  Metadata Sequence No  1
  VG Access              read/write
  VG Status              resizable
  MAX LV                 0
  Cur LV                 0
  Open LV                0
  Max PV                 0
  Cur PV                 1
  Act PV                 1
  VG Size                1020.00 MiB
  PE Size                4.00 MiB
  Total PE               255
  Alloc PE / Size       0 / 0
  Free  PE / Size       255 / 1020.00 MiB
  VG UUID               KrzkCo-QUFs-quJm-Z6pM-qMh0-ZchJ-c677c2

Creating the Logical Volumes and File Systems

Now that the volume group has been created, you can start creating one or more logical volumes (LVs) from it. This procedure is slightly more complicated than the creation of physical volumes or volume groups because there are more choices to be made. While creating the logical volume, you must specify a volume name and a size.

The volume size can be specified as an absolute value using the -L option. Use, for instance, -L 5G to create an LVM volume with a 5-GiB size. Alternatively, you can use relative sizes with the -l option. For instance, use -l 50%FREE to use half of all available disk space. You can also use the -l option to specify the number of extents that you want the logical volume to be. You’ll further need to specify the name of the volume group that the logical volume is assigned to, and optionally (but highly recommended), you can use -n to specify the name of the logical volume. For instance, use lvcreate -n lvdata -l 100 vgdata to create a logical volume with the name lvdata and a size of 100 extents and add that to the vgdata volume group. Once the logical volume has been created, you can use the mkfs utility to create a file system on top of it.

Understanding LVM Device Naming

Now that the logical volume has been created, you can start using it. To do this, you need to know the device name. LVM volume device names can be addressed in multiple ways. The simple method is to address the device as /dev/vgname/lvname. So, for example, if you have created a volume with the name lvdata, which gets its available disk space from the vgdata volume group, the device name would be /dev/vgdata/lvdata, which is in fact a symbolic link to the device mapper name (which is explained next).

For naming LVM volumes, another system plays a role: device mapper. The device mapper (abbreviated as dm) is a generic interface that the Linux kernel uses to address storage devices. The device mapper is used by multiple device types, such as LVM volumes, but also by software RAID and advanced network storage devices such as multipath devices.

Device mapper devices are generated on detection and use names that are generated while booting, like /dev/dm-0 and /dev/dm-1. To make these devices easier to access, the device mapper creates symbolic links in the /dev/mapper directory that point to these meaningless device names. The symbolic links follow the naming structure /dev/mapper/vgname-lvname.

So, the device /dev/vgdata/lvdata would also be known as /dev/mapper/vgdata-lvdata. When working with LVM logical volumes, you can use either of these device names. Example 15-7 shows an overview of the different LVM device names as provided by the device mapper. In Exercise 15-2, you learn how to create a volume group and logical volumes.

Example 15-7 LVM Device Name Overview

[root@server1 ~]# ls -l /dev/mapper/vgdata-lvdata /dev/vgdata/lvdata
lrwxrwxrwx. 1 root root 7 Sep 16 11:34 /dev/mapper/vgdata-lvdata ->
  ../dm-2
lrwxrwxrwx. 1 root root 7 Sep 16 11:34 /dev/vgdata/lvdata -> ../dm-2

Table 15-2 summarizes the relevant commands for creating logical volumes.

Resizing Volume Groups

The most important feature of LVM flexibility lies in the fact that it is so easy to resize the volume groups and the logical volumes that are using disk space from the volume groups. The vgextend command is used to add storage to a volume group, and the vgreduce command is used to take physical volumes out of a volume group. The procedure to add storage to a volume group is relatively easy:

1. Make sure that a physical volume or device is available to be added to the volume group.

2. Use vgextend to extend the volume group. The new disk space will show immediately in the volume group.

After extending a volume group, you can use the vgs command to verify that a physical volume has been added to the volume group. In Example 15-8, you can see that the vgdata VG contains two physical volumes, as indicated in the #PV column.

Example 15-8 Verifying VG Resize Operations with vgs

[root@server1 ~]# vgs
  VG      #PV  #LV  #SN   Attr     VSize        VFree
  centos    1    2    0   wz--n-   <19.00g      0
  vgdata    2    1    0   wz--n-   1020.00m     512.00m

Resizing Logical Volumes and File Systems

Like volume groups can be extended with the vgextend command, logical volumes can be extended with the lvextend command. This command has a very useful option -r to take care of extending the file systems on the logical volume at the same time; it is recommended to use this option and not the alternative approach that separately extends the logical volumes and the file systems on top of the logical volumes. Most file system resizing operations can be done online if the file system needs to be extended without unmounting it.

To grow the logical volume size, use lvextend or lvresize, followed by the -r option to resize the file system used on it. Then specify the size you want the resized volume to be. The easiest and most intuitive way to do that is by using -L followed by a + sign and the amount of disk space you want to add, as in lvresize -L +1G -r /dev/vgdata/lvdata. An alternative way to resize the logical volume is by using the -l option. This option is followed either by the number of extents that are added to the logical volume or by the absolute or relative percentage of extents in the volume group that will be used. You can, for example, use the following commands to resize the logical volume:

• lvresize -r -l 75%VG /dev/vgdata/lvdata Resizes the logical volume so that it will take 75% of the total disk space in the volume group. Notice that if currently the logical volume is using more than 75% of the volume group disk space, this command will try to reduce the logical volume size!

• lvresize -r -l +75%VG /dev/vgdata/lvdata Tries to add 75% of the total size of the volume group to the logical volume. This will work only if currently at least 75% of the volume group is unused. (Notice the difference with the previous command.)

• lvresize -r -l +75%FREE /dev/vgdata/lvdata Adds 75% of all free disk space to the logical volume.

• lvresize -r -l 75%FREE /dev/vgdata/lvdata Resizes the logical volume to a total size that equals 75% of the amount of free disk space, which may result in an attempt to reduce the logical volume size. (Notice the difference with the previous command.)

A logical extent is the logical building block used when creating logical volumes, and it maps to a physical extent, the size of which can be specified when creating a volume group. All resize operations need to match complete logical extents. You will sometimes notice that the resize size is rounded up or down to the logical extent size. You can also specify the number of logical extents that need to be added or removed directly by using the -l option with the lvresize command.

As you can see, resizing a logical volume has many options, and you need to take care to use the right options because it is easy to make a mistake! In Exercise 15-3, you learn how to resize logical volumes and the file systems used on top of them.

Reducing Volume Groups

If a volume group consists of multiple PVs, a PV can be removed from the VG if the remaining PVs have sufficient free space to allocate the extents it currently uses. This procedure will not work if the remaining PVs are fully used already. Removing a PV from a VG is a two-step procedure. First, use pvmove to move used extents from the PV that you want to remove to any of the remaining volumes. Next, use vgreduce to complete the PV removal. In Exercise 15-4 you can practice this.

Understanding Stratis Architecture

The lowest layer in the Stratis architecture is the pool. From a functional perspective, the Stratis pool is comparable to an LVM volume group. A pool represents all the available storage and consists of one or more storage devices, which in a Stratis environment are referred to as blockdev. These block devices may not be thin provisioned at the underlying hardware level. Stratis creates a /dev/stratis/poolname directory for each pool.

From the Stratis pool, XFS file systems are created. Note that Stratis only works with XFS, and the XFS file system it uses is integrated with the Stratis volume. When a file system is created, no size is specified, and each file system can grow up to the size of all the available storage space in the pool. Stratis file systems are always thin provisioned. The thin volume automatically grows as more data is added to the file system.

Creating Stratis Storage

Creating Stratis volumes is a multistep process. This section provides a high-level overview, and then Exercise 15-5 in the following section guides you through the procedure. You start by creating a pool. Once the pool has been added, you can create file systems from it. Before you begin, make sure that the block devices you’re going to use in Stratis have a minimal size of 5 GiB. Each Stratis file system occupies a minimum of 527 MiB of disk space, even if no data has been copied to the file system.

1. Install the Stratis software using dnf by installing the stratis-cli and stratisd packages.

2. Start and enable the user-space daemon, using systemctl enable --now stratisd.

3. Once the daemon is running, use the stratis pool create command to create the pool that you want to work with. For instance, use stratis pool create mypool /dev/sde to create a pool that is based on the block device /dev/sdd. You can add additional block devices later, using stratis pool add-data poolname blockdevname, as in stratis pool add-data mypool /dev/sde.

4. Once you have created the pool, add a file system using stratis fs create poolname fsname.

5. To verify that all was created successfully, use the stratis fs list command.

6. After creating the file system, you can mount it. To mount a Stratis file system through /etc/fstab, you must use the UUID; using the device name is not supported. Also, when mounting the Stratis volume through /etc/fstab, include the mount option x-systemd.requires=stratisd.service to ensure that the Systemd waits to activate this device until the stratisd service is loaded. Without this option you won’t be able to boot your system anymore.

Managing Stratis

After creating the Stratis file system, you can perform several different management tasks. To start with, you can dynamically extend the pool, using stratis pool add-data. Also, you need to monitor Stratis volumes using Stratis-specific tools, as the traditional Linux tools cannot handle the thin-provisioned volumes. The following commands are available:

• stratis blockdev: Shows information about all block devices that are used for Stratis.

• stratis pool: Gives information about Stratis pools. Note in particular the Physical Used parameter, which should not come too close to the Physical Size parameter.

• stratis filesystem: Enables you to monitor individual file systems.

Another Stratis feature that you may want to manage is the snapshot. A snapshot contains the state of the file system at the moment the snapshot was created. After creation, the snapshot can be modified. It’s also good to know that the snapshot and its origin are not linked, which allows the snapshot to live longer than the file system it was created from. This is fundamentally different from, for instance, LVM snapshots, which cannot stay alive if the volume they are linked to is removed.

In Exercise 15-5, you set up an environment with Stratis volumes.