5 LVM Configuration #

An LVM logical volume can optionally be thinly provisioned. Thin provisioning allows you to create logical volumes with sizes that overbook the available free space. You create a thin pool that contains unused space reserved for use with an arbitrary number of thin volumes. A thin volume is created as a sparse volume and space is allocated from a thin pool as needed. The thin pool can be expanded dynamically when needed for cost-effective allocation of storage space. Thinly provisioned volumes also support snapshots which can be managed with Snapper—see Chapter 7, System Recovery and Snapshot Management with Snapper for more information.

To set up a thinly provisioned logical volume, proceed as described in Procedure 5.1, “Setting Up a Logical Volume”. When it comes to choosing the volume type, do not choose Normal Volume, but rather Thin Volume or Thin Pool.

Important: Thinly Provisioned Volumes in a Cluster

To use thinly provisioned volumes in a cluster, the thin pool and the thin volumes that use it must be managed in a single cluster resource. This allows the thin volumes and thin pool to always be mounted exclusively on the same node.

A logical volume can be created with several mirrors. LVM ensures that data written to an underlying physical volume is mirrored onto a different physical volume. Thus even though a physical volume crashes, you can still access the data on the logical volume. LVM also keeps a log file to manage the synchronization process. The log contains information about which volume regions are currently undergoing synchronization with mirrors. By default the log is stored on disk and if possible on a different disk than are the mirrors. But you may specify a different location for the log, for example volatile memory.

Currently there are two types of mirror implementation available: "normal" (non-raid) mirror logical volumes and raid1 logical volumes.

After you create mirrored logical volumes, you can perform standard operations with mirrored logical volumes like activating, extending, and removing.

lvcreate -L 500G -m 2 -n lv1 vg1

lvcreate --type raid1 -m 1 -L 1G -n lv1 vg1

The lvresize, lvextend, and lvreduce commands are used to resize logical volumes. See the man pages for each of these commands for syntax and options information. To extend an LV there must be enough unallocated space available on the VG.

The recommended way to grow or shrink a logical volume is to use the YaST Partitioner. When using YaST, the size of the file system in the volume will automatically be adjusted, too.

LVs can be extended or shrunk manually while they are being used, but this may not be true for a file system on them. Extending or shrinking the LV does not automatically modify the size of file systems in the volume. You must use a different command to grow the file system afterward. For information about resizing file systems, see Chapter 2, Resizing File Systems.

Ensure that you use the right sequence when manually resizing an LV:

To extend the size of a logical volume:

For example, to extend an LV with a (mounted and active) Btrfs on it by 10 GB:

sudo lvextend −L +10G /dev/LOCAL/DATA
sudo btrfs filesystem resize +10G /dev/LOCAL/DATA

To shrink the size of a logical volume:

For example, to shrink an LV with a Btrfs on it by 5 GB:

sudo btrfs filesystem resize -size 5G /dev/LOCAL/DATA
sudo lvreduce /dev/LOCAL/DATA

Tip: Resizing the Volume and the File System with a Single Command

Starting with SUSE Linux Enterprise Server 12 SP1, lvextend, lvresize, and lvreduce support the option --resizefs which will not only change the size of the volume, but will also resize the file system. Therefore the examples for lvextend and lvreduce shown above can alternatively be run as follows:

sudo lvextend --resizefs −L +10G /dev/LOCAL/DATA
sudo lvreduce  --resizefs -L -5G /dev/LOCAL/DATA

Note that the --resizefs is supported for the following file systems: ext2/3/4, reiserfs, Btrfs, XFS. Resizing Btrfs with this option is currently only available on SUSE Linux Enterprise Server, since it is not yet accepted upstream.

Most LVM commands require an accurate view of the LVM metadata stored on the disk devices in the system. With the current LVM design, if this information is not available, LVM must scan all the physical disk devices in the system. This requires a significant amount of I/O operations in systems that have many disks. In case a disk fails to respond, LVM commands might run into a timeout while waiting for the disk.

Dynamic aggregation of LVM metadata via lvmetad provides a solution for this problem. The purpose of the lvmetad daemon is to eliminate the need for this scanning by dynamically aggregating metadata information each time the status of a device changes. These events are signaled to lvmetad by udev rules. If the daemon is not running, LVM performs a scan as it normally would do.

This feature is enabled by default. In case it is disabled on your system, proceed as follows to enable it:

LVM supports the use of fast block devices (such as an SSD device) as write-back or write-through caches for large slower block devices. The cache logical volume type uses a small and fast LV to improve the performance of a large and slow LV.

To set up LVM caching, you need to create two logical volumes on the caching device. A large one is used for the caching itself, a smaller volume is used to store the caching metadata. These two volumes need to be part of the same volume group as the original volume. When these volumes are created, they need to be converted into a cache pool which needs to be attached to the original volume:

For more information on LVM caching, see the lvmcache(7) man page.

After you tag the LVM2 storage objects, you can use the tags in commands to accomplish the following tasks:

A tag can be used in place of any command line LVM object reference that accepts:

Replacing the object name with a tag is not supported everywhere yet. After the arguments are expanded, duplicate arguments in a list are resolved by removing the duplicate arguments, and retaining the first instance of each argument.

Wherever there might be ambiguity of argument type, you must prefix a tag with the commercial at sign (@) character, such as @mytag. Elsewhere, using the “@” prefix is optional.

Consider the following requirements when using tags with LVM:

The following sections show example configurations for certain use cases.

tags {
   # Enable hostname tags
   hosttags = 1
}

tags {

   tag1 { }
      # Tag does not require a match to be set.

   tag2 {
      # If no exact match, tag is not set.
      host_list = [ "hostname1", "hostname2" ]
   }
}

  activation {
      volume_list = [ "vg1/lvol0", "@database" ]
  }

You can set up a simple host name activation control by enabling the hostname_tags option in the /etc/lvm/lvm.conf file. Use the same file on every machine in a cluster so that it is a global setting.

The examples in this section demonstrate two methods to accomplish the following: