Master Linux Disk Partitioning: From lsblk to LVM (Part 10)

1. lsblk – Inspect Disks

Before any operation, run lsblk to list all block devices, their sizes, types and mount points. Example output shows a 500 GB sda with a boot partition and an LVM group, and an empty 1 TB sdb. Useful flags: lsblk -f (filesystems), lsblk -l (list view), and lsblk -o NAME,SIZE,TYPE,FSTYPE,MOUNTPOINT for custom columns.

2. fdisk – Classic MBR Partitioning

fdisk

works on virtually every Linux system and supports MBR tables (max 2 TB, up to four primary partitions). After fdisk /dev/sdb you enter an interactive prompt where m shows help, p prints the table, n creates a partition, d deletes, t changes type, w writes changes (only w actually writes to disk), and q quits without saving.

Example: create a 500 GB primary partition on a new /dev/sdb, then run partprobe /dev/sdb so the kernel rereads the table.

3. parted – GPT for Large Disks

For disks larger than 2 TB, use GPT with parted. Basic interactive usage:

parted /dev/sdc
(parted) mklabel gpt        # create GPT label
(parted) mkpart primary ext4 0% 100%   # one partition filling the disk
(parted) print
(parted) quit

Non‑interactive (script‑friendly) form:

parted /dev/sdc mklabel gpt --script
parted /dev/sdc mkpart primary ext4 0% 100% --script

Tip: even for <2 TB disks you can use parted + GPT to avoid later migration.

4. Formatting – ext4 vs xfs

After partitioning, format the partition with a filesystem. Common commands: mkfs.ext4 /dev/sdb1 (stable, good for small files) mkfs.xfs /dev/sdb1 (high performance for large files, recommended by Oracle and Red Hat).

Guideline: use ext4 for general purpose, xfs for big‑data or database workloads. Verify the filesystem UUID with blkid /dev/sdb1.

5. mount/umount – Attach Filesystems

Mount a formatted partition:

mkdir /data
mount /dev/sdb1 /data

Common mount options (passed with -o): noatime, nodiratime, ro, rw, defaults. Unmount with umount /data (note the spelling). If the device is busy, identify processes with lsof /data or force release with fuser -km /data.

6. df – Check Free Space

Use df -h to display human‑readable disk usage. To find partitions over 90 % used, pipe through:

df -h | grep -E "(9[0-9]%|100%)"

7. du – Directory Usage

du -sh .

shows the total size of the current directory (e.g., 2.3G .). To locate the largest sub‑directories: du -h --max-depth=1 /var | sort -hr | head -10 Note the difference: du reports actual disk blocks used, while ls -lh shows logical file size.

8. dd – Low‑level Copying

Syntax: dd if=INPUT of=OUTPUT bs=BLOCK_SIZE count=COUNT. Common uses:

Backup a partition:

dd if=/dev/sdb1 of=/backup/sdb1_backup.img bs=4M status=progress

Create a bootable USB:

dd if=ubuntu-22.04.iso of=/dev/sdc bs=4M status=progress && sync

Wipe a disk: dd if=/dev/zero of=/dev/sdb bs=4M status=progress Warning: dd does not ask for confirmation; mixing up if and of can permanently erase a system disk.

9. LVM – Flexible Volume Management

LVM abstracts physical volumes (PV) into volume groups (VG) and logical volumes (LV), allowing online resizing. Core concepts:

PV – actual disk partitions

VG – pool of PVs

LV – allocatable space, similar to a traditional partition

Example creation:

# 1. pvcreate /dev/sdb1 /dev/sdc1
# 2. vgcreate vg_data /dev/sdb1 /dev/sdc1
# 3. lvcreate -L 500G -n lv_data vg_data
# 4. mkfs.ext4 /dev/vg_data/lv_data
mkdir /data
mount /dev/vg_data/lv_data /data

To extend when /data fills up:

# lvextend -L +100G /dev/vg_data/lv_data
# resize2fs /dev/vg_data/lv_data   # for ext4
# xfs_growfs /data                 # for xfs

Additional PVs can be added with pvcreate /dev/sdd1 and vgextend vg_data /dev/sdd1. LVM adds flexibility but introduces an extra abstraction layer and cannot be used for /boot (GRUB cannot read LVM).

10. /etc/fstab – Persistent Mounts

Temporary mounts disappear after reboot; to make them permanent, add entries to /etc/fstab:

<device>    <mountpoint>    <fstype>    <options>    <dump>    <fsck>

Typical line using UUID: UUID=a1b2c3d4-... /data ext4 defaults,noatime 0 2 After editing, verify with mount -a. A malformed fstab can prevent the system from booting, as illustrated by a real‑world case where an extra space broke the boot process.

Summary

Key workflow for a new server: run lsblk to discover disks, use fdisk for ≤2 TB or parted for >2 TB, format with ext4 (general) or xfs (large files), mount and record the mount in /etc/fstab, verify with mount -a, and employ LVM when frequent resizing is needed. Handle dd with extreme care to avoid catastrophic data loss.