df / du

# -----------------------------------------------
#  df (Disk Free) — Check free space across all filesystems
# -----------------------------------------------

# df -h
#   → Shows disk usage for all mount points in human-readable units
#   → Sizes are displayed in GB, MB, etc.
#   Example: df -h

# df -i
#   → Shows inode usage instead of block usage
#   → When inodes are exhausted, no new files can be created
#   Example: df -i

# df -h {mount_point}
#   → Shows only the specified mount point
#   Example: df -h /var

# -----------------------------------------------
#  du (Disk Usage) — Check usage by directory
# -----------------------------------------------

# du -sh {directory}
#   → Shows the total size of the specified directory in human-readable units
#   → -s: summary only (does not expand subdirectories)
#   → -h: human-readable units (K/M/G)
#   Example: du -sh /var/log

# du -h --max-depth=1 {directory}
#   → Shows the size of each immediate subdirectory under the given path
#   → --max-depth=1: expands only one level deep
#   Example: du -h --max-depth=1 /var

# du -hs * | sort -rh
#   → Lists all entries in the current directory sorted by size, largest first
#   → sort -r: reverse order (largest first)  sort -h: correctly compares human-readable sizes
#   Example: cd /var && du -hs * | sort -rh

# -----------------------------------------------
#  ncdu — Interactive TUI disk investigation tool
# -----------------------------------------------

# ncdu {directory}
#   → Displays a directory tree in TUI and lets you drill down with the keyboard
#   → Install: sudo apt install ncdu (Debian/Ubuntu)
#   →          sudo dnf install ncdu (RHEL/AlmaLinux)
#   Example: ncdu /

# -----------------------------------------------
#  find — Detect large files
# -----------------------------------------------

# find {directory} -size +{size} -type f
#   → Lists files exceeding the specified size
#   → +100M: over 100 MB  +1G: over 1 GB
#   Example: find / -size +100M -type f 2>/dev/null

Command	Description
df -h	Shows free space for all mount points in human-readable units (K/M/G). Any mount point where Use% exceeds 90% warrants attention.
df -i	Shows inode usage. When a filesystem holds an enormous number of files, inodes can run out even when block space is still available, preventing new files from being created.
df -h /var	Shows free space for a specific mount point only.
du -sh {directory}	Displays the total size of the specified directory in a single line. Useful for a quick size estimate.
du -h --max-depth=1 {directory}	Expands one level of subdirectories under the given path and shows their sizes. Useful for identifying which subdirectory is taking up the most space.
du -hs * | sort -rh	Lists entries in the current directory sorted by size, largest first. Lets you quickly spot the largest directories.
ncdu {directory}	Displays a directory tree in TUI and lets you drill down with the arrow keys. You can also delete files with the d key.
find / -size +100M -type f 2>/dev/null	Searches for files larger than 100 MB. Useful for finding old backups or bloated log files.
find / -name "*.log" -size +50M 2>/dev/null	Searches for log files larger than 50 MB. Helpful for finding logs that have grown out of control due to a missing logrotate configuration.

Investigating the cause of a full disk

# -----------------------------------------------
#  Steps to diagnose disk exhaustion
#  (when / reaches 98% on Kiryu Kazuma's server)
# -----------------------------------------------

# Step 1: Check which mount point is running low
df -h

# Step 2: Find which top-level directory is using the most space
du -h --max-depth=1 /

# Step 3: If /var looks large, drill down further
du -h --max-depth=1 /var

# Step 4: If /var/log looks suspicious, inspect the log directory
du -hs /var/log/* | sort -rh

# Step 5: List files over 100 MB to find clues
find /var -size +100M -type f 2>/dev/null

# Step 6: Use ncdu to visually browse and delete unnecessary files
# (move with j/k or arrow keys, delete with d)
ncdu /var

Run the following command:

$ df -h
Filesystem      Size  Used Avail Use% Mounted on
/dev/sda1        50G   48G  980M  98% /
/dev/sda2       100G   12G   88G  12% /home
tmpfs           7.8G     0  7.8G   0% /dev/shm
$ du -h --max-depth=1 /var
4.0K	/var/backups
2.1G	/var/log
890M	/var/lib
44G	/var/lib/kiryu_db_backup
4.0K	/var/cache
46G	/var
$ find /var -size +100M -type f 2>/dev/null
/var/lib/kiryu_db_backup/dump_20260101.sql.gz
/var/lib/kiryu_db_backup/dump_20260201.sql.gz
/var/lib/kiryu_db_backup/dump_20260301.sql.gz
/var/log/nginx/majima_access.log.1

Diagnosing inode exhaustion

# -----------------------------------------------
#  Check whether inodes are exhausted
#  (when Majima Goro's server shows "No space left on device"
#   but df -h shows plenty of free space)
# -----------------------------------------------

# Check inode usage
# IUse% close to 100% indicates inode exhaustion
df -i

# Find directories consuming the most inodes
# (identifies directories with a very large number of files)
find / -xdev -printf '%h\n' 2>/dev/null | sort | uniq -c | sort -rn | head -20

Run the following command:

$ df -i
Filesystem      Inodes  IUsed   IFree IUse% Mounted on
/dev/sda1      3276800 3276795       5  100% /
/dev/sda2      6553600  124032 6429568    2% /home
$ find / -xdev -printf '%h\n' 2>/dev/null | sort | uniq -c | sort -rn | head -5
 892341 /var/spool/majima_mail_queue
  84210 /tmp/nishiki_sessions
   4821 /var/log
    312 /usr/lib
    201 /etc

Investigating interactively with ncdu

# -----------------------------------------------
#  Installing and using ncdu
# -----------------------------------------------

# Install on Debian/Ubuntu
sudo apt install ncdu

# Install on RHEL/AlmaLinux
sudo dnf install ncdu

# Start scanning from / (may take a while)
ncdu /

# Scan only /home for a quicker investigation
ncdu /home

# Save the scan result to a file for later review
# (useful when you want to minimize load on a production server)
ncdu -o /tmp/tatsuya_disk_report.json /
ncdu -f /tmp/tatsuya_disk_report.json

Run the following command:

$ ncdu /home
ncdu 1.19 ~ Use the arrow keys to navigate, press ? for help
--- /home -----------------------------------------------------------------------
   18.2 GiB [##########] /kiryu
    8.7 GiB [####      ] /majima
    3.1 GiB [#         ] /nishiki
  512.0 MiB [          ] /date_true
   48.3 MiB [          ] /shinada_work
(↓ to move, → or Enter to drill down, d to delete)

df and du are the essential first tools for troubleshooting disk space issues. The standard investigation flow is: use df -h to get a bird's-eye view of all filesystems and identify mount points with a high Use%, then drill down with du -h --max-depth=1 to narrow down the cause. If you get "No space left on device" even though block space looks fine, check df -i for inode exhaustion — mail queues and temporary files can accumulate in huge numbers, draining inodes before disk blocks run out. For tracking down large files, find -size is effective at quickly surfacing old backups or bloated, unrotated logs. To prevent log bloat at the root level, review your logrotate configuration. Having a solid understanding of the Linux directory structure (FHS) also helps you decide which directories to prioritize when interpreting du results.