LVM

From ParabolaWiki
Jump to: navigation, search
Summary
This article will provide an example of how to install and configure Parabola with Logical Volume Manager (LVM).
Required software
[ Software]
Related
Software RAID and LVM
System Encryption with LUKS

1 Introduction

See the Wikipedia article on this subject for more information: Logical Volume Manager (Linux)

LVM (Logical Volume Management) makes use of the device-mapper feature of the Linux-libre kernel to provide a system of partitions that is independent of the underlying disk's layout. With LVM you can abstract your storage space and have "virtual partitions" which makes it easier to extend and shrink partitions (subject to the filesystem you use allowing this) and add/remove partitions without worrying about whether you have enough contiguous space on a particular disk, without getting caught up in the problems of fdisking a disk that is in use (and wondering whether the kernel is using the old or new partition table) and without having to move other partition out of the way. This is strictly an ease-of-management issue: it does not provide any addition security. However, it sits nicely with the other two technologies we are using.

Note that LVM is not used for the boot partition, because of the bootloader problem.

The basic building blocks of LVM are:

  • Physical volume (PV): Partition on hard disk (or even hard disk itself or loopback file) on which you can have volume groups. It has a special header and is divided into physical extents. Think of physical volumes as big building blocks which can be used to build your hard drive.
  • Volume group (VG): Group of physical volumes that are used as storage volume (as one disk). They contain logical volumes. Think of volume groups as hard drives.
  • Logical volume (LV): A "virtual/logical partition" that resides in a volume group and is composed of physical extents. Think of logical volumes as normal partitions.
  • Physical extent (PE): A small part of a disk (usually 4MB) that can be assigned to a logical Volume. Think of physical extents as parts of disks that can be allocated to any partition.

With LVM you can more easily handle your partitions (logical volumes) than normal hard drive partitions. For example, you can:

  • Use any number of disks as one big disk(VG)
  • Have partitions(LV) stretched over several disks (they can be as big as all of your disk storage together)
  • Resize/create/delete partitions(LV) and disks(VG) as you like (it does not depend on position of the logical volumes within volume groups as with normal partitions)
  • Resize/create/delete partitions(LV) and disks(VG) online (filesystems on them still need to be resized, but some support online resizing)
  • Name your disks(VG) and partitions(LV) as you like
  • Create small partitions(LV) and resize them "dynamically" as they get more filled (growing must be still done by hand, but you can do it online with some filesystems)
  • ...

Example:

Physical disks
                
  Disk1 (/dev/sda):
     _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
    |Partition1 50GB (Physical volume) |Partition2 80GB (Physical volume)     |
    |/dev/sda1                         |/dev/sda2                             |
    |_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ |_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ |
                                  
  Disk2 (/dev/sdb):
     _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
    |Partition1 120GB (Physical volume)                 |
    |/dev/sdb1                                          |
    | _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ _ _|
LVM logical volumes

  Volume Group1 (/dev/MyStorage/ = /dev/sda1 + /dev/sda2 + /dev/sdb1):
     _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
    |Logical volume1 15GB  |Logical volume2 35GB      |Logical volume3 200GB               |
    |/dev/MyStorage/rootvol|/dev/MyStorage/homevol    |/dev/MyStorage/mediavol             |
    |_ _ _ _ _ _ _ _ _ _ _ |_ _ _ _ _ _ _ _ _ _ _ _ _ |_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ |

To sum it all up: With LVM you can use all your storage space as one big disk (volume group) and have more flexibility over your partitions (logical volumes).

2 Advantages

Here are some things you can do with LVM that you cannot (or cannot do easily) with just mdadm, MBR partitions, GPT partitions, parted/gparted and a file-level tool like rsync.

  1. Online/live partition resizing
  2. No need for an extended partition (not relevant for GPT)
  3. Resize partitions regardless of their order on disk (no need to ensure surrounding available space)
  4. Online/live migration of partitions being used by services without having to restart services

These can be very helpful in a server situation, desktop less so, but you must decide if the features are worth the abstraction.

3 Installation

Before doing anything we need to load the appropriate module:

# modprobe dm-mod

If you already have Parabola installed and you just want to add/try a partition with LVM, jump to partition disks.

3.1 Installing Parabola on LVM

Prior to running Parabola install scripts (/libre/setup) to install Parabola, you need to partition your disk with cfdisk (or any other tool of your liking). Because grub legacy (grub with version less than 1.0) cannot boot from LVM logical volumes you cannot have /boot in LVM, so you need to create a boot partition. 100MB should be enough. The other solution would be to use lilo or grub with version 1.95 or newer.

3.2 Partition disks

Next you need to create a partition for LVM. Its filesystem type should be 'Linux LVM', so use a partition id 0x8e (filesystem type: 8e). You need to create only one LVM partition on each disk you want to use with LVM. Your logical volumes will reside inside these partitions so size them accordingly. If you will use only LVM and no other external partitions, use all the free space on each disk.

Warning: /boot cannot reside in an LVM partition when using GRUB versions < 1.95, which does not support LVM. If you want /boot to reside in a LVM partition, use GRUB2.
Tip: All LVM partitions on all disks can be configured to appear as one big disk.

3.3 Create Physical volumes

Now you need to initialize these partitions so they can be used by LVM. Use fdisk -l to find out which partitions have filesystem type 'Linux LVM' and create a physical volume on them:

# pvcreate /dev/sda1

Substitute /dev/sda1 with all your partitions to create physical volumes on all of them. This command creates a header on each partition so it can be used for LVM. You can track created physical volumes with:

# pvdisplay

3.4 Create Volume group(s)

Next step is to create a volume group on this physical volume. First you need to create a volume group on one of the new partitions and then add to it all other physical volumes you want to have in it:

# vgcreate VolGroup00 /dev/sda1
# vgextend VolGroup00 /dev/sdb1

Also you can use any other name you like instead of VolGroup00 for a volume group when creating it. You can track how your volume group grows with:

# vgdisplay
Note: You can create more than one volume group if you need to, but then you will not have all your storage presented as one disk.

3.5 Create Logical Volumes

Now we need to create logical volumes on this volume group. You create a logical volume with the next command by giving the name of a new logical volume, its size, and the volume group it will live on:

# lvcreate -L 10G VolGroup00 -n lvolhome

This will create a logical volume that you can access later with /dev/mapper/Volgroup00-lvolhome or /dev/VolGroup00/lvolhome. Same as with the volume groups, you can use any name you want for your logical volume when creating it.

To create swap on a logical volume, an additional argument is needed:

# lvcreate -C y -L 10G VolGroup00 -n lvolswap

The -C y is used to create a contiguous partition, which means that your swap space does not get partitioned over one or more disks nor over non-contiguous physical extents.

If you want to fill all the free space left on a volume group, use the next command:

# lvcreate -l +100%FREE VolGroup00 -n lvolmedia

You can track created logical volumes with:

# lvdisplay
Note: You may need to load the device-mapper kernel module (modprobe dm-mod) for the above commands to succeed.
Tip: You can start out with relatively small logical volumes and expand them later if needed. For simplicity, leave some free space in the volume group so there is room for expansion.

3.6 Create filesystem and mount logical volumes

Your logical volumes should now be located in /dev/mapper/ and /dev/YourVolumeGroupName. If you cannot find them, use the next commands to bring up the module for creating device nodes and to make volume groups available:

# modprobe dm-mod
# vgscan
# vgchange -ay

Now you can create filesystems on logical volumes and mount them as normal partitions (if you are installing Parabola, refer to mounting the partitions for additional details):

# mkfs.ext4 /dev/mapper/VolGroup00-lvolhome
# mount /dev/mapper/VolGroup00-lvolhome /home

If you are installing Parabola, start /libre/setup, go to Prepare Hard Drive directly to step 3 Set Filesystem Mountpoints and read the Important section below before proceeding with installation!

3.7 Important

There are just a few things you need to be careful with while using/installing Parabola with LVM.

3.7.1 Set Filesystem Mountpoints

  • When choosing mountpoints, just select your newly created logical volumes (use: /dev/mapper/Volgroup00-lvolhome).
    Do NOT select the actual partitions on which logical volumes were created (do not use: /dev/sda2).

3.7.2 Configure System

  • Make sure you change USELVM="no" to USELVM="yes" in /etc/rc.conf. The rc.sysinit script that parses the USELVM variable entry will accept either yes or YES, however it will not accept mixed case [This is fixed in latest release].
  • Make sure that lvm2 is in the HOOKS section of /etc/mkinitcpio.conf just before the filesystems so that your kernel will find LVM volumes at boot time.
  • If your root filesystem ( "/" ) is put onto a logical volume, make sure regenerate kernel image based on above modified /etc/mkinitcpio.conf by using below command so that bootloader can find your root during booting
     cd /boot 
     mkinitcpio -p linux-libre 
  • Make sure your bootloader uses the right volumes for root.
For GRUB, edit /boot/grub/menu.lst:
     ...
     (0) Parabola GNU/Linux-libre
     title  Parabola GNU/Linux-libre
     root   (hd0,0)
     kernel /vmlinuz-linux-libre root=/dev/mapper/VolGroup00-lvolroot resume=/dev/mapper/VolGroup00-lvolswap ro
     initrd /initramfs-linux-libre.img
     ...
For SYSLINUX, edit /boot/syslinux/syslinux.cfg:
     ...
     # (0) Parabola GNU/Linux-libre
     LABEL parabola
       MENU LABEL Parabola
       LINUX ../vmlinuz-linux-libre
       APPEND root=/dev/mapper/VolGroup00-lvolroot ro
       INITRD ../initramfs-linux-libre.img
     ...
For LILO, edit /etc/lilo.conf:
     ...
     image=/boot/vmlinuz-linux-libre
       label=parabola
       append="root=/dev/mapper/VolGroup00-lvolroot resume=/dev/mapper/VolGroup00-lvolswap ro"
       initrd=/boot/initramfs-linux-libre.img
     ...

4 Configuration

4.1 Grow logical volume

To grow a logical volume you first need to grow the logical volume and then the filesystem to use the newly created free space. Let us say we have a logical volume of 15GB with ext3 on it and we want to grow it to 20G. We need to do the following steps:

# lvextend -L 20G VolGroup00/lvolhome (or lvresize -L +5G VolGroup00/lvolhome)
# resize2fs /dev/VolGroup00/lvolhome

You may use lvresize instead of lvextend.

If you want to fill all the free space on a volume group, use the next command:

# lvextend -l +100%FREE VolGroup00/lvolhome
Warning: Not all filesystems support growing without loss of data and/or growing online.
Note: If you do not resize your filesystem, you will still have a volume with the same size as before (volume will be bigger but partly unused).

4.2 Shrink logical volume

Because your filesystem is probably as big as the logical volume it resides on, you need to shrink the filesystem first and then shrink the logical volume. Depending on your filesystem, you may need to unmount it first. Let us say we have a logical volume of 15GB with ext3 on it and we want to shrink it to 10G. We need to do the following steps:

# resize2fs /dev/VolGroup00/lvolhome 9G
# lvreduce -L 10G VolGroup00/lvolhome (or lvresize -L -5G VolGroup00/lvolhome)
# resize2fs /dev/VolGroup00/lvolhome

Here we shrunk the filesystem more than needed so that when we shrunk the logical volume we did not accidentally cut off the end of the filesystem. After that we normally grow the filesystem to fill all free space left on logical volume. You may use lvresize instead of lvreduce.

Warning:
  • Do not reduce the filesystem size to less than the amount of space occupied by data or you risk data loss.
  • Not all filesystems support shrinking without loss of data and/or shrinking online.
Note: It is better to reduce the filesystem to a smaller size than the logical volume, so that after resizing the logical volume, we do not accidentally cut off some data from the end of the filesystem.

4.3 Remove logical volume

Warning: Before you remove a logical volume, make sure to move all data that you want to keep somewhere else, otherwise it will be lost!

First, find out the name of the logical volume you want to remove. You can get a list of all logical volumes installed on the system with:

# lvs

Next, look up the mountpoint for your chosen logical volume...:

$ df -h

... and unmount it:

# umount /your_mountpoint

Finally, remove the logical volume:

# lvremove /dev/yourVG/yourLV

Confirm by typing y and you are done.

Do not forget, to update /etc/fstab!

You can verify the removal of your logical volume by typing "lvs" as root again (see first step of this section).

4.4 Add partition to a volume group

To add a partition to your volume group you must first make its type 'Linux LVM' (for example with cfdisk). Then you need to create a physical volume on it and extend the volume group over it:

# pvcreate /dev/sdb1
# vgextend VolGroup00 /dev/sdb1

Now you have free space in your volume group that can be used by logical volumes in this group.

Tip: You can add partitions from any disks to volume groups.

4.5 Remove partition from a volume group

All of the data on that partition needs to be moved to another partition. Fortunately, LVM makes this easy:

# pvmove /dev/sdb1

If you want to have the data on a specific physical volume, specify that as the second argument to pvmove:

# pvmove /dev/sdb1 /dev/sdf1

Then the physical volume needs to be removed from the volume group:

# vgreduce myVg /dev/sdb1

Or remove all empty physical volumes:

# vgreduce --all vg0

And lastly, if you want to use the partition for something else, and want to avoid LVM thinking that the partition is a physical volume:

# pvremove /dev/sdb1

4.6 Snapshots

4.6.1 Introduction

LVM allows you to take a snapshot of your system in a much more efficient way than a traditional backup. It does this efficiently by using a COW (copy-on-write) policy. The initial snapshot you take simply contains hard-links to the inodes of your actual data. So long as your data remains unchanged, the snapshot merely contains its inode pointers and not the data itself. Whenever you modify a file or directory that the snapshot points to, LVM automatically clones the data, the old copy referenced by the snapshot, and the new copy referenced by your active system. Thus, you can snapshot a system with 35GB of data using just 2GB of free space so long as you modify less than 2GB (on both the original and snapshot).

4.6.2 Configuration

You create snapshot logical volumes just like normal ones.

# lvcreate --size 100M --snapshot --name snap01 /dev/mapper/vg0-pv

With that volume, you may modify less than 100M of data, before the snapshot volume fills up.

It is important to have the dm-snapshot module listed in the MODULES variable of /etc/mkinitcpio.conf, otherwise the system will not boot. If you do this on an already installed system, make sure to rebuild the image with

# mkinitcpio -g /boot/initramfs-linux-libre.img

Todo: scripts to automate snapshots of root before updates, to rollback... updating menu.lst to boot snapshots (separate article?)

snapshots are primarily used to provide a frozen copy of a filesystem to make backups; a backup taking two hours provides a more consistent image of the filesystem than directly backing up the partition.

5 Troubleshooting

5.1 LVM commands do not work

  • Load proper module:
# modprobe dm_mod
  • Try preceding commands with lvm like this:
# lvm pvdisplay

5.2 Set Filesystem Mountpoints page does not show logical volumes

If you are installing on a system where there is an existing volume group, you may find that even after doing "modprobe dm-mod" you do not see the list of logical volumes.

In this case, you may also need to do:

# vgscan

or:

# vgchange -ay <volgroup>

in order to activate the volume group and make the logical volumes available.

5.3 Receiving Input/Output Errors after plugging in a removable device with LVM partitions

Symptoms:

~$ sudo vgscan
 Reading all physical volumes.  This may take a while...
 /dev/backupdrive1/backup: read failed after 0 of 4096 at 319836585984: Input/output error
 /dev/backupdrive1/backup: read failed after 0 of 4096 at 319836643328: Input/output error
 /dev/backupdrive1/backup: read failed after 0 of 4096 at 0: Input/output error
 /dev/backupdrive1/backup: read failed after 0 of 4096 at 4096: Input/output error
 Found volume group "backupdrive1" using metadata type lvm2
 Found volume group "networkdrive" using metadata type lvm2

Cause:

Removing an external LVM drive without deactivating the volume group(s) first. Before you disconnect, make sure to:
# vgchange -an <volume group name>

Fix: (assuming you already tried to activate the volume group with vgchange -ay <vg>, and are receiving the Input/output errors

# vgchange -an <volume group name>
Unplug the external drive and wait a few minutes
# vgscan
# vgchange -ay <volume group name>

6 Additional resources

7 Acknowledgement

This wiki article is based on ArchWiki. We may have removed non-FSDG bits from it.