Managing storage volumes

Add storage volumes

Things to consider before you add storage:

• How much storage do you need
• What name did the system assign the new device? /dev/sd[a..z], /dev/vd[a..z]
  • For example, the disk/device is named /dev/sdb and the first partition is named /dev/sdb1
• How to format the disk - what filesystem?
• Where to mount the disk

Find devices

# --- Options to locate attached devices ---#
fdisk -l            # view device info
dmesg --follow      # watch kernel ring buffer as you attach new device
dmesg | tail        # view end of kernel ring buffer for new device
lsblk               # list block devices

Format and partition storage

After you find your device, create a partition with fdisk:

• Linux stores partition configurations in partition tables. There are two types of partition tables:
  • Master Boot Record (MBR): Legacy version
    • fdisk might refer to this as DOS (Disk Operating System)
    • Only lets you create up to 4 partitions
    • Can only partition up to 2TB of disk
  • GUID Partition Table (GPT): New standard
    • Create up to 128 partitions

# --- Create partition on disk --- #
fdisk /dev/sdb          # 1. Run fdisk on the device
p                       # 2. Verify that you are working on the correct device
g                       # 3. Create a GPT partition table
n                       # 4. Create a new partition
ENTER                   # 5. Accept default partition number
ENTER                   # 6. Accept default for first sector of partition
+10G                    # 7. Set the partition size - this is 10 gigabytes (set the last partition sector)
p                       # 8. View changes
w                       # 9. Write changes to the partition table

# --- Format new partition --- #
mkfs.ext4 /dev/sdb1

# --- Delete a partition --- #
fdisk /dev/sdb          # 1. Run fdisk on the device
p                       # 2. Verify that you are working on the correct device
d                       # 3. Delete command
[1..]                   # 4. Select the partition number (if more than one)

fdisk

fdisk -l            # view device info
fdisk <device>      # start partitioning on <device>

Mount and unmount volumes

When you mount a volume, you attach a storage device or network share to a local directory on your server:

• Local dir must be empty.
• Volume is another way to say “filesystem” or “partition”
• Filesystem Hierarchy Standard recommends these dirs for mountpoints:
  • /mnt for mounted fs you plan to use regularly - hard drives, network-attached storage
  • /media: removable media - flash drives, external hard drives
  • To mount a USB drive, create a usb1/ dir. Ex: /media/usb1/

# --- Mount a volume --- #
mkdir /media/usb1                       # 1. Create empty dir to mount
df -h                                   # 2. View currently mounted filesystems
mount /dev/sdb1 /media/usb1             # 3. Mount volume to new empty directory
mount /dev/sdb1 -t ext4 /media/usb1     #    If there is an error about fs type
df -h                                   # 4. Confirm volume was mounted correctly

# --- Unmount a volume --- #
umount /media/usb1                      # 1. Unmount (no "n") to disconnect the volume from the fs
                                        #    Volume cannot be in use, or returns an error

mount

Attaches a storage device or network share to a local directory on your server:

mount           # list all mounted filesystems

/etc/fstab

At boot, the system looks at this file to mount your main fs, swap, and any other file:

• Each partition is ID’d with a universally unique identifier (UUID)
  • Can list regular partition name–/dev/sdb1– but UUID is preferred bc partiton names change depending on where you mount them or the order in which they are attached to the system
  • UUID does not change unless you reformat your volume

Column descriptions

Options included in defaults:

• rw: Device is mounted as read/write
• exec: Allow files on volume to be executed as scripts
• auto: Automatically mount device at boot time or when you run sudo mount -a
• nouser: Only root can mount the fs
• async: Output to the device is asynchronous

Other options:

• ro: Device is mounted as read-only
• noexec: Prevent storing executable scripts on mounted volume
• noauto: Must mount manually. Cannot mount at boot or with sudo mount -a. This is good for drives used for backups that don’t need to be mounted all the time.
• users: Give regular users access without sudo or using root privileges

Adding volume to /etc/fstab

blkid /dev/sdb1                     # 1. Find UUID of volume
mkdir /mnt/extra_storage_fstab      # 2. Create mountpoint dir

# --------------------------------- # 3. Add info to /etc/fstag
# Extra storage
UUID=8609c662-378d-4df0-be90-f35d1e1fdccc /mnt/extra_storage_fstab ext4 defaults 0 0

sudo mount -a                       # 4. Mount without reboot
df -h                               # 5. Verify mounted volume

# --- Manually mount w/fstab --- #
# read/write, manually mount
UUID=8609c662-378d-4df0-be90-f35d1e1fdccc /mnt/ext_disk ext4 rw,noauto 0 0

mount /mtn/ext_disk                  # just list dest path for mount

blkid

List UUIDs of all volumes known to your system:

• might have to run as root

blkid                   # list all blkid
blkid <partition>       # list UUID for <partition>

RAID

Lets you join multiple disks in a variety of configurations to prevent data loss:

• Has different levels, where the higher the level, the more disks that can fail before you have a data loss problem
• RAID level 1: two hard disks always have the same data, one can fail
  • If a disk fails, you can rebuild the array
• RAID level 5: same as level 1 but more disks (?)
• RAID level 6: two disks can fail before
• NOT A BACKUP SOLUTION - you can still lose data if disks are fried or stolen

Backups

Backups duplicate data:

• Should be offsite or not in the same physical location as your data
• Should be resilient and let you get your data back up and running quickly
  • You should test your backups regularly
• Have three layers - NAS, cloud, offsiet, mirroring data
• Consider encrypting data, wherever it is

LVM

Linux Volume Manager.

You can resize your filesystem/partitions without rebooting or shutting down your server:

• ALWAYS use LVM on storage volumes for a virtual server, when possible
• For virtual (cloud) disks, you can grow a filesystem without a server reboot
  • Likely require reboot if you need to attach physical disk to physical server
• hot-plugging is adding or removing hardware components to a running server without having to shut it down or reboot it

Terminology

Volume group

Namespace that includes all physical and logical volumes for an implementation of LVM.

• Use a common naming convention, such as vg-<volume-name>.

Physical volume

Physical or virtual hard disk that is a member of a volume group.

Logical volume

A flexible, resizable partition that acts as a single unit whose size can span multiple physical volumes.

Implementation

Create a volume group, assign physical disk space to that group, then split up the physical volumes into logical volumes:

• LVM expects a block device, so make sure there are no partitions or partition signatures on the disk
  • sudo wipefs -a /dev/sdb deletes the partition signatures
• When you add physical volumes to a volume group with vgextend, do not assign the PVs to a logical volume until the space is needed
• View LVs, PVs, and VGs with xxdisplay command. For example, lvdisplay, pvdisplay, vgdisplay

apt install lvm2                    # Install package
lvs                                 # Monitor logical volume size
# --- Set up Logical Volume --- #

pvcreate /dev/sdb                               # 1. Designate each disk as a physical volume
pvdisplay                                       # 2. View phyical volumes and confirm changes
vgcreate vg-test /dev/sdb                       # 3. Create the volume group and assign a physical disk
vgdisplay                                       # 4. View volume groups and confirm changes
lvcreate -n myvol1 -L 5g vg-test                # 5. Create 5G logical volume named myvol1 and assign to vg-test
lvdisplay                                       # 6. View logical volumes and confirm changes

# --- Format Logical Volume --- #

lvdisplay                                       # 1. Get device name (LV Path)
mkfs.ext4 /dev/vg-test/myvol1                   # 2. Format logical volume
mkdir -p /mnt/lvm/myvol1                        # 3. Create empty dir for mountpoint
mount /dev/vg-test/myvol1 /mnt/lvm/myvol1/      # 4. Mount the logical volume
df -h                                           # 5. Verify volume was mounted

# --- Resizing Logical Volume --- #

lvextend -n /dev/vg-test/myvol1 -l +100%FREE    # 1. Extend LV to take up remaining PV disk space
df -h                                           # 2. Get name of filesystem on LV
resize2fs /dev/mapper/vg--test-myvol1           # 3. Resize filesystem to span entire LV

# --- Add PVs to volume group --- #

vgextend vg-test /dev/sdc                       # 1. Add PV to volume group
lvextend -L+10g /dev/vg-test/myvol1             # 2. Add 10G from the new PV to the logical volume
resize2fs /dev/vg-test/myvol1                   # 3. Resize the filesystem to span entire LV

# --- Removing LVMs --- #

umount /mnt/lvm/myvol1                          # 1. Unmount the logical volume
lvremove vg-test/myvol1                         # 2. Remove LV myvol1 from VG vg-test
vgremove vg-test                                # 3. Remove the volume group

LVM Snapshots

Captures a logical volume at a specific point in time and preserves it:

• You can mount a snapshot or use it to revert the snapshot in case of failure
  • Ex: you want to test how security updates will affect your server and might need to revert back
  • NOT A FORM OF BACKUP - not stored in different location, can become corrupt if disk space gets low
• Requires unallocated space in volume group
• Is a clone of the original LV
• Use case: Take a snapshot of the root filesystem before installing all available updates. Rollback if there is an issue, or delete if everything is OK.

lvcreate -s -n mysnapshot -L 4g vg-test/myvol1  # Create snapshot mysnapshot of LV myvol1
lvconvert --merge vg-test/mysnapshot            # Roll back to mysnapshot (must unmount first)
lvremove vg-test/mysnapshot                     # Delete the snapshot

wipefs

Deletes partition signatures from a disk:

wipefs -a /dev/sdb