Table of Contents
+ +NetBSD LVM is build on Linux lvm2tools and libdevmapper with device-mapper kernel driver + written for NetBSD.
+The LVM driver allows the user to manage available disk space. Disk space + from several disks, partitions can be added to “Volume Groups”, which is pool + available disk space for “Logical Partitions” + aka Logical Volumes.
+The steps required to setup a LVM are as follows:
+-
+
Install physical media
+Configure kernel support
+Configure system, install tools
+-
+
+ Optional step +
+Disklabel each volume member of the LVM
+
+ Initialize the LVM disk devices
+Create a volume group from initialized disks
+Create Logical volume from created Volume group
+Create a filesystem on the new LV device
+Mount the LV filesystem
+
This example features a LVM setup on NetBSD/i386 -current.
+
-
+
-
+
Volume Group
+The Volume Group is the highest level abstraction used within the LVM. It gathers + together a collection of Logical Volumes and Physical Volumes into one administrative + unit.
+
+ -
+
Physical Volume
+A physical volume is typically a hard disk, though it may well just be a device that + 'looks' like a hard disk (eg. a software raid device).
+
+ -
+
Logical Volume
+The equivalent of a disk partition in a non-LVM system. The LV is visible as a + standard block device; as such the LV can contain a file system (eg. /home).
+
+ -
+
Physical Extends
+Each physical volume is divided chunks of data, known as physical extents, these + extents have the same size as the logical extents for the volume group.
+
+ -
+
Logical Extends
+Each logical volume is split into chunks of data, known as logical extents. The extent + size is the same for all logical volumes in the volume group.
+
+ -
+
Physical Extends mapping
+There are couple of general strategies used for mapping “LE's” to + “PE's”.
++-
+
-
+
+ Linear Mapping +
+This mapping lineary will assing range of PE's to LE's.
+For example it can map 100 PE's from PV 1 to LV 1 and + another 100 PE's from PV 0.
+
+ -
+
+ Stripe Mapping +
+will interleave the chunks of the logical extents across a number of physical + volumes.
+++Warning
+Be aware that Stripe mapping is not implemented yet in NetBSD device-mapper + driver.
+
+
+
+
+ -
+
-
+
Snapshots
+A wonderful facility provided by LVM is 'snapshots'. This allows the administrator to + create a new block device which presents an exact copy of a logical volume, frozen at some + point in time. Typically this would be used when some batch processing, a backup for + instance, needs to be performed on the logical volume, but you don't want to halt a live + system that is changing the data. When the snapshot device has been finished with the + system administrator can just remove the device. This facility does require that the + snapshot be made at a time when the data on the logical volume is in a consistent state.
+++Warning
+Snapshot feature is not fully implemented in NetBSD and should not be used in + production.
+
+
This step is at your own discretion, depending on your platform and the hardware at your + disposal. LVM can be used with disklabel partitions or even with standard + partitions created with fdisk.
+From my DMESG:
+Disk #1: + probe(esp0:0:0): max sync rate 10.00MB/s + sd0 at scsibus0 target 0 lun 0: <SEAGATE, ST32430N SUN2.1G, 0444> SCSI2 0/direct fixed + sd0: 2049 MB, 3992 cyl, 9 head, 116 sec, 512 bytes/sect x 4197405 sectors + + Disk #2 + probe(esp0:1:0): max sync rate 10.00MB/s + sd1 at scsibus0 target 1 lun 0: <SEAGATE, ST32430N SUN2.1G, 0444> SCSI2 0/direct fixed + sd1: 2049 MB, 3992 cyl, 9 head, 116 sec, 512 bytes/sect x 4197405 sectors + + Disk #3 + probe(esp0:2:0): max sync rate 10.00MB/s + sd2 at scsibus0 target 2 lun 0: <SEAGATE, ST11200N SUN1.05, 9500> SCSI2 0/direct fixed + sd2: 1005 MB, 1872 cyl, 15 head, 73 sec, 512 bytes/sect x 2059140 sectors + + Disk #4 + probe(esp0:3:0): max sync rate 10.00MB/s + sd3 at scsibus0 target 3 lun 0: <SEAGATE, ST11200N SUN1.05, 8808 > SCSI2 0 + sd3: 1005 MB, 1872 cyl, 15 head, 73 sec, 512 bytes/sect x 2059140 sectors+
The following kernel configuration directive is needed to provide LVM device support. It + is not enabled in the GENERIC kernel:
+Note
+This driver is only available in haad-dm branch. Therefore you have + to update your src/sys directory to haad-dm branch.
+pseudo-device dm # device-mapper device+
For using LVM you have to install lvm2tools and libdevmapper to NetBSD system. These tools + are not enabled as default.
+To enable building of LVM tools set MKLVM=yes in mk.conf file.
Note
+These tools are only available in haad-dm branch. Therefore you + have to update your src/external/gpl2 directory to haad-dm branch.
+ After installation of lvm2tools and libdevmapper /etc/lvm directory
+ should be created. For debuging purposes it is good to copy
+ lvm2tools/dist/doc/example.conf there as lvm.conf.
Each physical volume disk in LVM will need a special file system established. In this + example, I will need to disklabel:
+
+ /dev/rsd0d
+ /dev/rsd1d
+ /dev/rsd2d
+ /dev/rsd3d
+
Note
+Always remember to disklabel the character device, not the block device, in
+ /dev/r{s,w}d*
Note
+On all platforms except i386 where d partition is used for this,
+ the c slice is symbolic of the entire NetBSD partition and is
+ reserved.
You will probably want to remove any pre-existing disklabels on the physical voleme disks + in the LVM. This can be accomplished in one of two ways with the dd(1) command:
++#dd if=/dev/zero of=/dev/rsd0d bs=8k count=1+#dd if=/dev/zero of=/dev/rsd1d bs=8k count=1+#dd if=/dev/zero of=/dev/rsd2d bs=8k count=1+#dd if=/dev/zero of=/dev/rsd3d bs=8k count=1
If your port uses a MBR (Master Boot Record) to partition the disks so that the NetBSD + partitions are only part of the overall disk, and other OSs like Windows or Linux use other + parts, you can void the MBR and all partitions on disk by using the command:
++#dd if=/dev/zero of=/dev/rsd0d bs=8k count=1+#dd if=/dev/zero of=/dev/rsd1d bs=8k count=1+#dd if=/dev/zero of=/dev/rsd2d bs=8k count=1+#dd if=/dev/zero of=/dev/rsd3d bs=8k count=1
This will make all data on the entire disk inaccessible. Note that the entire disk is
+ slice d on i386 (and some other ports), and c
+ elsewhere (e.g. on sparc). See the “kern.rawpartition” sysctl - "3" means "d",
+ "2" means "c".
The default disklabel for the disk will look similar to this:
++#disklabel -r sd0+ [...snip...] + bytes/sector: 512 + sectors/track: 63 + tracks/cylinder: 16 + sectors/cylinder: 1008 + cylinders: 207 + total sectors: 208896 + rpm: 3600 + interleave: 1 + trackskew: 0 + cylinderskew: 0 + headswitch: 0 # microseconds + track-to-track seek: 0 # microseconds + drivedata: 0 + + 4 partitions: + # size offset fstype [fsize bsize cpg/sgs] + a: 208896 0 4.2BSD 0 0 0 # (Cyl. 0 - 207*) + d: 208896 0 unused 0 0 # (Cyl. 0 - 207*)
You will need to create one “slice” on the NetBSD partition of the disk + that consumes the entire partition. The slice must begin at least two sectors after end of + disklabel part of disk. On i386 it is sector “63”. Therefore, the + “size” value should be “total sectors” minus 2x + “sectors”. Edit your disklabel accordingly:
++#disklabel -e sd0
Note
+The offset of a slice of type “4.2BSD” must be a multiple of the + “sectors” value.
+Note
+Be sure to export EDITOR=[path to your favorite editor] before + editing the disklabels.
+Note
+The slice must be fstype 4.2BSD.
Because there will only be one slice on this partition, you can recycle the
+ d slice (normally reserved for symbolic uses). Change your disklabel to the
+ following:
3 partitions: + # size offset fstype [fsize bsize cpg] + d: 4197403 65 4.2BSD # (Cyl. 1 - 4020*)+
Optionally you can setup a slice other than d to use, simply adjust
+ accordingly below:
3 partitions: + # size offset fstype [fsize bsize cpg] + a: 4197403 65 4.2BSD # (Cyl. 1 - 4020*) + c: 4197405 0 unused 1024 8192 # (Cyl. 0 - 4020*)+
Be sure to write the label when you have completed. Disklabel will object to your + disklabel and prompt you to re-edit if it does not pass its sanity checks.
+Once all disks are properly labeled, you will need to create physical volume on them. + Every partition/disk added to LVM must have physical volume header on start + of it. All informations, like Volume group where Physical volume belongs are stored in this + header.
+++#lvm pvcreate /dev/rwd1[ad]+
Status of physical volume can be viewed with pvdisplay command.
+++#lvm pvdisplay+
Once all disks are properly labeled with physical volume header, volume group must be + created from them. Volume Group is pool of PE's from which administrator can create Logical + Volumes “partitions”.
+++#lvm vgcreate vg0 /dev/rwd1[ad]+
-
+
vg0 is name of Volume Group
+/dev/rwd1[ad] is Physical Volume
+
Volume group can be later extended/reduced with vgextend and vgreduce commands. These + commands adds physical volumes to VG.
+++#lvm vgextend vg0 /dev/rwd1[ad]+ +#lvm vgreduce vg0 /dev/rwd1[ad]+
Status of Volume group can be viewed with vgdisplay command.
+++#lvm vgdisplay vg0+
Once Volume Group was created administrator can create “logical partitions” + volumes.
+++#lvm lvcreate -L 20M -n lv1 vg0+
-
+
vg0 is name of Volume Group
+-L 20M is size of Logical Volume
+-n lv1 is name of Logical Volume
+
Logical Volume can be later extended/reduced with lvextend and lvreduce commands.
+++#lvm lvextend -L+20M /dev/vg0/lv1+ +#lvm lvreduce -L-20M /dev/vg0/lv1+
Note
+To shrink lv partition you have to shrink filesystem before and NetBSD ffs2 doesn't + support it now.
+Status of Logical Volume can be viewed with lvdisplay command.
+++#lvm vgdisplay vg0/lv1+