Ok, don't let this scare you off, but LVM comes with a lot of jargon which you should understand lest you endanger your filesystems.
We start at the bottom, more or less.
You should take the word 'physical' with a grain of salt, though we will initially assume it to be a simple hard disk, or a partition. Examples, /dev/hda, /dev/hda6, /dev/sda. You can turn any consecutive number of blocks on a block device into a ...
A PV is nothing more than a physical medium with some administrative data added to it - once you have added this, LVM will recognise it as a holder of ...
Physical Extents are like really big blocks, often with a size of megabytes. PEs can be assigned to a...
A VG is made up of a number of Physical Extents (which may have come from multiple Physical Volumes or hard drives). While it may be tempting to think of a VG as being made up of several hard drives (/dev/hda and /dev/sda for example), it's more accurate to say that it contains PEs which are provided by these hard drives.
>From this Volume Group, PEs can be assigned to a ...
Yes, we're finally getting somewhere. A Logical Volume is the end result of our work, and it's there that we store our information. This is equivalent to the historic idea of partitions.
As with a regular partition, on this Logical Volume you would typically build a ...
This filesystem is whatever you want it to be: the standard ext2, ReiserFS, NWFS, XFS, JFX, NTFS, etc... To the linux kernel, there is no difference between a regular partition and a Logical Volume.
I've attempted some ASCII art which may help you visualise this.
A Physical Volume, containing Physical Extents: +-----[ Physical Volume ]------+ | PE | PE | PE | PE | PE | PE | +------------------------------+ A Volume Group, containing 2 Physical Volumes (PVs) with 6 Physical Extents: +------[ Volume Group ]-----------------+ | +--[PV]--------+ +--[PV]---------+ | | | PE | PE | PE | | PE | PE | PE | | | +--------------+ +---------------+ | +---------------------------------------+ We now further expand this: +------[ Volume Group ]-----------------+ | +--[PV]--------+ +--[PV]---------+ | | | PE | PE | PE | | PE | PE | PE | | | +--+---+---+---+ +-+----+----+---+ | | | | | +-----/ | | | | | | | | | | | | +-+---+---+-+ +----+----+--+ | | | Logical | | Logical | | | | Volume | | Volume | | | | | | | | | | /home | | /var | | | +-----------+ +------------+ | +---------------------------------------+
This shows us two filesystems, spanning two disks. The /home filesystem contains 4 Physical Extents, the /var filesystem 2.
bert hubert is writing a tool to represent LVM more visually, a screenshot is provided. Looks better than the ASCII art.
Ok, this stuff is hard to assimilate ('We are LVM of Borg...'), so here is a very annotated example of creating a Logical Volume. Do NOT paste this example onto your console because you WILL destroy data, unless it happens that on your computer /dev/hda3 and /dev/hdb2 aren't used.
When in doubt, view the ASCIIgram above.
You should first set the partition types of /dev/hda3 and /dev/hdb2 to 0x8e, which is 'Linux LVM'. Please note that your version of fdisk may not yet know this type, so it will be listed as 'Unknown':
# fdisk /dev/hda Command (m for help): p Disk /dev/hda: 255 heads, 63 sectors, 623 cylinders Units = cylinders of 16065 * 512 bytes Device Boot Start End Blocks Id System /dev/hda1 1 2 16033+ 83 Linux /dev/hda2 3 600 4803435 83 Linux /dev/hda3 601 607 56227+ 83 Linux /dev/hda4 608 614 56227+ 83 Linux Command (m for help): t Partition number (1-4): 3 Hex code (type L to list codes): 8e Command (m for help): p Disk /dev/hda: 255 heads, 63 sectors, 623 cylinders Units = cylinders of 16065 * 512 bytes Device Boot Start End Blocks Id System /dev/hda1 1 2 16033+ 83 Linux /dev/hda2 3 600 4803435 83 Linux /dev/hda3 601 607 56227+ 8e Unknown /dev/hda4 608 614 56227+ 83 Linux Command (m for help): w
We do the same for /dev/hdb2, but we don't display it here. This is needed so that LVM is able to reconstruct things should you lose your configuration.
Now, this shouldn't be necessary, but some computers require a reboot at this point. So if the following examples don't work, try rebooting.
We then create our Physical Volumes, like this:
# pvcreate /dev/hda3 pvcreate -- physical volume "/dev/hda3" successfully created # pvcreate /dev/hdb2 pvcreate -- physical volume "/dev/hdb2" successfully created
We than add these two PVs to a Volume Group called 'test':
# vgcreate test /dev/hdb2 /dev/hda3 vgcreate -- INFO: using default physical extent size 4 MB vgcreate -- INFO: maximum logical volume size is 255.99 Gigabyte vgcreate -- doing automatic backup of volume group "test" vgcreate -- volume group "test" successfully created and activated
So we now have an empty Volume Group, let's examine it a bit:
Lots of data here - most of it should be understandable by now. We see that there are no Logical Volumes defined, so we should work to remedy that. We try to generate a 50 megabyte volume called 'HOWTO' in the Volume Group 'test':# vgdisplay -v test --- Volume group --- VG Name test VG Access read/write VG Status available/resizable VG # 0 MAX LV 256 Cur LV 0 Open LV 0 MAX LV Size 255.99 GB Max PV 256 Cur PV 2 Act PV 2 VG Size 184 MB PE Size 4 MB Total PE 46 Alloc PE / Size 0 / 0 Free PE / Size 46 / 184 MB --- No logical volumes defined in test --- --- Physical volumes --- PV Name (#) /dev/hda3 (2) PV Status available / allocatable Total PE / Free PE 13 / 13 PV Name (#) /dev/hdb2 (1) PV Status available / allocatable Total PE / Free PE 33 / 33
# lvcreate -L 50M -n HOWTO test lvcreate -- rounding up size to physical extent boundary "52 MB" lvcreate -- doing automatic backup of "test" lvcreate -- logical volume "/dev/test/HOWTO" successfully created
Ok, we're nearly there, let's make a filesystem:
# mke2fs /dev/test/HOWTO mke2fs 1.18, 11-Nov-1999 for EXT2 FS 0.5b, 95/08/09 Filesystem label= OS type: Linux Block size=1024 (log=0) Fragment size=1024 (log=0) 13328 inodes, 53248 blocks 2662 blocks (5.00%) reserved for the super user First data block=1 7 block groups 8192 blocks per group, 8192 fragments per group 1904 inodes per group Superblock backups stored on blocks: 8193, 24577, 40961 Writing inode tables: done Writing superblocks and filesystem accounting information: done # mount /dev/test/HOWTO /mnt # ls /mnt lost+found
And we're done! Let's review our Volume Group, because it should be filled up a bit by now:
# vgdisplay test -v --- Volume group --- VG Name test VG Access read/write VG Status available/resizable VG # 0 MAX LV 256 Cur LV 1 Open LV 1 MAX LV Size 255.99 GB Max PV 256 Cur PV 2 Act PV 2 VG Size 184 MB PE Size 4 MB Total PE 46 Alloc PE / Size 13 / 52 MB Free PE / Size 33 / 132 MB --- Logical volume --- LV Name /dev/test/HOWTO VG Name test LV Write Access read/write LV Status available LV # 1 # open 1 LV Size 52 MB Current LE 13 Allocated LE 13 Allocation next free Read ahead sectors 120 Block device 58:0 --- Physical volumes --- PV Name (#) /dev/hda3 (2) PV Status available / allocatable Total PE / Free PE 13 / 13 PV Name (#) /dev/hdb2 (1) PV Status available / allocatable Total PE / Free PE 33 / 20
Well, it is. /dev/hda3 is completely unused, but /dev/hdb2 has 13 Physical Extents in use.
As with all decent operating systems, Linux is divided in two parts: kernel space and user space. Userspace is sometimes called userland, which would also be a good name for a theme park, 'Userland'.
Discovery, creation and modification of things pertaining to Logical Volume Management is done in user space, and then communicated to the kernel. Once a volume group or logical volume is reported to the kernel, it is called 'Active'. Certain changes can only be performed when an entity is active, others only when it is not.