Ubuntu Tinkering Notes

Introduction

#

Ubuntu, as a popular Linux distribution, offers better ecosystem support compared to other Linux distros. The most notable advantage is that when you encounter issues, you’re more likely to find tutorials and solutions for Ubuntu.

This article primarily focuses on the GUI-based personal edition of Ubuntu. For server-specific Ubuntu systems, operations depend on your actual business needs. The article Cloud Service Deployment can serve as a reference.

Ubuntu Installation

#

The installation was done a long time ago, so no detailed records exist. If needed, please search for keywords like “installing Ubuntu on a portable hard drive.” Here’s a relatively recent guide: Installing Ubuntu on a Portable Hard Drive.

The installation process is no longer traceable. The current setup involves installing Ubuntu on a portable hard drive, allowing it to be used on-the-go.

To use it, simply plug in the hard drive before powering on the computer. Quickly press a specific key to enter the BIOS boot menu, set the priority to the highest, save, and exit to boot into Ubuntu.

To switch back to Windows, just unplug the hard drive and power on normally—no additional steps are required.

Interface Customization

#

I personally prefer an Apple-style interface, so I specifically chose an Apple-inspired theme: WhiteSur.

The installation process is straightforward:

git clone https://github.com/vinceliuice/WhiteSur-gtk-theme.git --depth=1  

cd WhiteSur-gtk-theme  

./install.sh # Run the installation script  

For detailed configuration, refer to the instructions on the GitHub page.

As for updates, the official guide doesn’t specify, presumably assuming users already know:

git pull # Fetch the latest code  

./install.sh # Re-run the installation script  

Text Editing

#

Text editing here specifically refers to editing text in the command line. Editing text in a dedicated text editor is so straightforward and simple that it doesn’t require learning. And in most cases, we don’t need to edit text in the command line.

However, there are times when you have to edit text in the command line: servers often lack a GUI😢 Even though you can run a GUI editor on a server using VS Code plugins, launching such a “heavyweight” editor just to modify a configuration file feels like overkill🤔

If you simply want to create a file and then edit it with another editor, one simple command will suffice:

# Just touch it and it will appear😉
touch file_name.md

nano

#

This command-line editor is pre-installed in Ubuntu systems, so there’s no need for a separate download.

Editing a file is very simple:

nano file_name

Once inside nano, there are a few lines of prompts that highlight the relevant shortcuts; just follow the instructions. It’s a very simple command-line text editor.

vim

#

I’ve heard about this “legendary ancient” editor for a long time and learned about its many advantages: keeping your fingers on the keyboard helps with thought flow, once you get used to it you can’t quit, it’s minimalist and highly customizable, all cool programmers use vim…

After several attempts, I came to the conclusion: vim is great, but I’m really lazy😢 I really don’t want to spend weeks getting used to all kinds of key combinations.

Here I only document the most commonly used operations, suitable for those who occasionally use vim for lightweight editing. After mastering these operations, you can claim that you know how to use vim😋

i # Enter insert mode

<ESC> # Return to normal mode

: # Enter command mode

:q! # Quit without saving

:wq # Save and quit

y6w # Copy 6 words

p # Paste

dd # Delete a line

dw # Delete a word

neovim

#

Due to various unavoidable reasons, I occasionally have to use a terminal IDE for project development 😢. However, the traditional Vim editor is just too “hardcore”. I recently heard that Neovim is quite user-friendly and decided to give it a try.

Neovim is a modern reimplementation of Vim, short for Neo (New) Vim. Its goal is to bring modern features to the classic Vim editor, including:

• A better plugin system (asynchronous processing, Lua support)
• Enhanced language server integration (LSP)
• More user-friendly configuration options (using Lua instead of VimScript)
• Better integration with IDE-like features (debugging, completion, navigation, etc.)

Directory Structure

#

Main configuration directory structure:

~/.config/nvim/
├── init.lua                 # Entry file, similar to VS Code's settings.json
├── lua/                     # Contains all your Lua configuration modules
│   ├── core/                # General settings, keymaps, custom functions, etc.
│   ├── plugins/             # Plugin configuration files (one file per plugin)
│   └── lazy.lua             # Initialization file for lazy.nvim plugin loader
├── after/                   # Additional settings executed after plugins are loaded
│   └── plugin/              # E.g., filetype autocommands
└── plugin/                  # Lua/Vim scripts that are auto-loaded

Configuration Solutions

#

Nowadays, many incredibly sleek terminal IDEs are actually powered by Neovim under the hood. These IDEs are essentially pre-configured setups crafted by Neovim enthusiasts—essentially the results of their meticulous tinkering. Some well-known examples include: LazyVim and NvChad.

These third-party configurations usually come with comprehensive documentation, so you can simply follow the official instructions to install them.

Fcitx 5

#

Main reference: Install and Configure Fcitx 5 Chinese Input Method on Ubuntu. I initially considered using Sogou Input Method, but the official installation process seemed overly complicated, and it also required installing Fcitx 5. So, I figured I might as well just use Fcitx 5 directly.

To be honest, I was reluctant to use Fcitx at first 🥲 because its interface is so “plain” that it’s hard to accept. I believe the author of the blog post above must have felt the same way 👆.

Using Windows Fonts

#

Approach: Copy font files from Windows to Ubuntu’s dedicated font directory, assign appropriate permissions, refresh Ubuntu’s font cache, and load the new fonts.

# Windows font directory: C:/Windows/Fonts  
sudo cp /mnt/C/Windows/Fonts/LXGWWenKai-Regular.ttf /usr/share/fonts/custom/LXGWWenKai-Regular.ttf  

# Grant permissions  
sudo chmod u+rwx /usr/share/fonts/custom/*  

# Navigate to the font directory  
cd /usr/share/fonts/custom/  

# Create font cache  
sudo mkfontscale  

# Refresh cache  
sudo fc-cache -fv  

Alternatively, you can download a new .ttf file from the web and copy it to the target directory. If you’re using a GUI-based Ubuntu system, you can simply double-click the font file to install it 🥰.

Setting Up VPN

#

I usually use Clash + airport to access external network resources.

First, of course, follow the airport tutorial to download Clash and obtain the configuration file. Then, a VPN is typically started using a command like this:

./clash -d .

However, this command occupies a terminal and requires manual input every time after booting, which is quite inconvenient.

Therefore, this command can be encapsulated into a system service, allowing it to be controlled and managed using systemctl. It can also be set to start automatically on boot.

Sample configuration file, which needs to be written to /etc/systemd/system/clash.service:

[Unit]
Description=Clash Service
After=network.target

[Service]
Type=simple
User=morethan
ExecStart=/home/morethan/Program/clash/clash -d /home/morethan/Program/clash
Restart=on-failure
RestartSec=5

[Install]
WantedBy=multi-user.target

Set it to start automatically on boot:

sudo systemctl daemon-reexec && sudo systemctl enable clash.service

Control it using systemctl:

# Start immediately
sudo systemctl start clash.service

# Stop immediately
sudo systemctl stop clash.service

Of course, you can also manually configure it in: System Settings --> Network --> Proxy.

SSH Reverse Proxy

#

We often need to use remote servers for code development work. However, networks of domestic remote servers have particularly tricky characteristics: they cannot connect to the external internet, or sometimes even cannot access the domestic network at all.

The reason I call it a ‘characteristic’ is mainly because I honestly have never seen a server that can fulfill both of the aforementioned network requirements 😭. I once lost code I had worked on for several months due to network issues, and it still pains me to think about it 😭.

So, after much reflection, I decided to thoroughly solve this thorny problem: henceforth, whenever I can connect to a remote server, I must ensure it can smoothly access the external network. As the saying goes, sharpening the axe won’t delay the job of cutting wood. Network issues are truly not something that can be overlooked.

The following content is primarily sourced from this blog post: Tutorial on Using SSH Reverse Tunnels to Connect Remote Computers to the Internet

The core command is as follows:

# Open a terminal and start the reverse proxy
ssh -R 7890:localhost:7890 username@remote_host

The meaning of this command is: open port 7890 on the remote computer and forward all data sent to this port to port 7890 on the local computer (i.e., your proxy server). After executing this command, a terminal session to the remote server will open. You can immediately test if the proxy is successful in this terminal:

https_proxy=http://localhost:7890 curl https://www.baidu.com

If a VPN happens to be running on port 7890 of your local computer, you can also try accessing resources on the external network:

https_proxy=http://localhost:7890 curl https://www.google.com

If everything goes well, you should now be able to access external network resources 😄.

However, I find the method above somewhat inelegant: every time you want to set up a reverse proxy for a server, you need to run this command, and it also occupies a terminal window.

Therefore, I encapsulated this functionality into a system service template, allowing control using commands similar to the following:

# Start the reverse proxy
sudo systemctl start ssh-reverse-proxy@A6000.service

# Check the proxy status
sudo systemctl status ssh-reverse-proxy@A6000.service

# Stop the reverse proxy
sudo systemctl stop ssh-reverse-proxy@A6000.service

The part after the @ is a variable parameter; you just need to fill in the alias of the target server you want to access. Of course, you need to define your server aliases in ~/.ssh/config. Also, for convenience, password-less login should be configured. See the next section for details: Password-less SSH Login

You can also wrap the lengthy command above into an alias; the specific name is a matter of personal preference, as long as it’s concise and understandable.

Here are the command operations for configuring the system-level service:

sudo vim /etc/systemd/system/ssh-reverse-proxy@.service

Write the following content (note, some essential parts need modification):

[Unit]
# The '%i' in Description will be replaced by the parameter you pass (the SSH alias)
Description=SSH Reverse Proxy to %i
After=network-online.target
Wants=network-online.target

[Service]
# [MUST MODIFY] Replace with your local username so the service can find your SSH keys
User=your_local_user

# The '%i' in ExecStart will also be replaced by the SSH alias
# This is the core enabling parameterization
ExecStart=/usr/bin/ssh -NT \
    -o ServerAliveInterval=60 \
    -o ExitOnForwardFailure=yes \
    -R 7890:localhost:7890 %i

# Automatic restart policy
Restart=always
RestartSec=10

[Install]
WantedBy=multi-user.target

Then reload the systemd configuration, and you can use the systemctl commands mentioned above to control this reverse proxy service 😄.

If you want to use Git, you’ll need to route SSH traffic through HTTPS using netcat. It might sound abstract, but the actual steps are quite straightforward. Simply add the following content to the ~/.ssh/config file on the server:

Host github.com gitlab.com
  # Match multiple hosts separated by spaces

  # [Core Configuration]
  # Use netcat (nc) to connect via an HTTP proxy
  # -x connect_address:port specifies the HTTP proxy address
  # -X 5 specifies the proxy protocol as SOCKS5 (if the proxy supports it)
  #  - or -
  # -X connect specifies the use of HTTP CONNECT (more universal)
  ProxyCommand nc -X connect -x localhost:7890 %h %p
  
  User git

Then, follow the instructions in GitHub-SSH to configure password-free push for GitHub.

Password-less SSH Login

#

For students without a local GPU, editing and running files directly on the server is very convenient, and the VSCode remote-ssh extension can also present a VSCode interface directly from the server. To avoid the pain of constantly entering passwords, you can perform the following operations to achieve password-less login 😄.

Assuming the local machine is Windows, SSH-related configuration files are generally stored in the C:\Users\<User_name>\.ssh folder. Achieving password-less login only requires working with the files in this directory.

1. Configure the config file. A sample configuration is as follows:

# Replace <User_name> with your local machine username
Host 3090
    HostName xx.xxx.xx.xx
    User morethan
    IdentityFile "C:\Users\<User_name>\.ssh\id_rsa"

2. Generate the authentication file id_rsa.pub (if not already present).

3. Locally create an authorized_keys file and write the contents of id_rsa.pub into this file.

4. Create a .ssh folder in the default directory on the server, and then copy the local authorized_keys file into it.

After that, when you try to log into the server using VSCode, you’ll find that password-less login is already set up.

Mounting Hard Drives

#

Since my Ubuntu system is installed on a portable hard drive, the main goal here is to access Windows partitions from Ubuntu. This section doesn’t cover detailed partition operations. For tasks like formatting partitions, refer to: How to Partition and Mount Disks in Ubuntu.

# View disks and partitions (sudo privileges required)  
sudo fdisk -l  

# Create a mount point (essentially creating a folder)  
# The subfolder under /mnt is named "E" because it’s intended to mount the E drive  
sudo mkdir /mnt/E  

# Mount the new partition directly  
sudo mount /dev/vdb /mnt/E  

# Set auto-mount at boot  
# Check the partition’s UUID  
sudo blkid  

# Edit the specific file  
vim /etc/fstab  

# Append to the end of the file  
UUID=xxxxxxxx /mnt/E ntfs defaults 0 2  

• Replace the UUID above with the output from blkid. Replace ntfs with the appropriate filesystem type (common types include ntfs and ext 4).
• defaults: This is a combination of default mount options, such as rw (read-write) and relatime (reduces inode access time updates).
• 0 and 2: These values control backup and filesystem check order. Typically, the first value is 0 (no backup), and the second is 1 or 2 (1 for the root filesystem, 2 for others).

Testing method:

# If no errors occur, the configuration is correct  
sudo mount -a  

Hard Drive Monitoring

#

Since my Ubuntu is installed on an external hard drive, monitoring the drive’s status is necessary for system security.

Let me explain why hard drive monitoring is needed with my personal experience. Because it’s an external drive, it occasionally gets accidentally bumped, causing the drive to lose power abruptly. After a forced shutdown and restart, the system might seem fine on the surface, but there’s a high chance the drive has already suffered permanent damage 😭 – specifically, what are known as “bad blocks.” By the time I realized it, the number of bad blocks had already reached 262.

Furthermore, if the number of bad blocks increases rapidly in a short period—for example, increasing by 10 within a week—it’s crucial to back up your data immediately and prepare to replace the drive.

Fortunately, hard drive monitoring is a very common requirement, so there’s a software called smartmontools available for this purpose. Run the following command to install it:

sudo apt update && sudo apt install smartmontools

Then, modify the configuration file to monitor the specific drive device:

sudo vim /etc/smartd.conf

# Comment out all existing content, and add the following line at the end:
/dev/sdb -a -d sat -o on -S on -I 194

Note: The content added above needs to be adjusted based on your actual situation, such as the drive name at the beginning.

Once configured, you can start the service directly. The monitoring program will then run automatically after boot:

# Start the service
sudo systemctl start smartd

# Check its status
sudo systemctl status smartd

To view the status, you can directly check the system logs:

# View all logs related to smartd
journalctl -u smartd

# View only messages of warning level and above
journalctl -u smartd -p warning

The following steps are optional and aim to redirect smartd related logs to a separate file. Since I have system auto-cleanup enabled, system logs are only retained for two days. However, hard drive monitoring typically needs to be assessed over longer periods, like months.

Modify the system logging service configuration file:

sudo vim /etc/rsyslog.d/60-smartd.conf

Add the following content:

if $programname == 'smartd' then /var/log/smartd.log
& stop

The specific meaning of this configuration is: Redirect all logs related to smartd to the /var/log/smartd.log file and prevent them from being written to the system log (to avoid duplication).

Then restart the services:

# Restart the logging service
sudo systemctl restart rsyslog

# Restart smartd
sudo systemctl restart smartd

# Verify smartd status
sudo systemctl status smartd

# Verify log redirection
cat /var/log/smartd.log

You’ll notice that the content in smartd.log is quite verbose. Therefore, you can create an alias to filter this information, showing only the important details:

alias checksmartd='grep -iE "Reallocated_Sector|warning|error|fail|changed" /var/log/smartd.log | grep -v "Offline Testing failed" | grep -v "ignoring -l error"'

After restarting your terminal, run:

checksmartd

It’s normal to see no output immediately after configuration. However, if bad blocks increase, relevant content will be displayed.

Creating Shortcuts

#

A common task: placing a quick link to a frequently used folder on the desktop for easy access.

# Place a link to /target/dir in the Desktop folder  
# Replace with your target directory  
ln -s /target/dir ~/Desktop  

# Test—if you can cd into it, it works  
cd ~/Desktop/dir  

System Information

#

When there is no UI, you need to use the command line to check relevant system information.

# Summary of system information
hostnamectl

# CPU details
lscpu

# Check CPU usage
top

# Partition usage
df -h

Configuring Git

#

One of the standout features of Linux is its extreme simplicity, which is why using the command line to manage Git is the preferred choice for Linux users 😃. Ubuntu comes with Git pre-installed, so there’s no need to install it separately. If you want to upgrade, follow these steps:

git --version # Check the Git version

sudo add-apt-repository ppa:git-core/ppa # Add the official repository

sudo apt update && sudo apt upgrade # If possible, proceed with the upgrade

GitHub-SSH

#

Of course, you can also use HTTPS directly, but the downside is that you’ll need to enter your password every time. Moreover, with GitHub’s increasing security measures, the password isn’t necessarily your account password but rather a dedicated token 🥲.

Such a cumbersome process is unbearable on Linux. I’d rather go through a tedious setup once than have to enter a long token every time.

This section is mainly referenced from: Configuring Git to Push by Default Without Entering Credentials (Ubuntu, SSH).

git config --global user.name 'xx' # Configure the global username

git config --global user.email 'xxx@qq.com' # Configure the global email account

# Generate an SSH key pair. Here, I choose to press Enter all the way through.
ssh-keygen -t rsa -b 4096 -C "your_email@example.com"

# Start the SSH agent and load the private key into the agent
eval "$(ssh-agent -s)"
ssh-add ~/.ssh/id_rsa

# View and copy the public key content
cat ~/.ssh/id_rsa.pub

# Add this new SSH key to your GitHub account.

# Change an existing HTTPS-linked repository to an SSH link
git remote rm origin
git remote add origin git@github.com:username/repository.git

Installing and Managing Software

#

Software installation on Ubuntu generally falls into the following categories:

1. Via the built-in Snap store
2. Via apt
3. Via .deb packages
4. Via curl

Different installation methods require different management approaches. curl installations are the most cumbersome to manage, while others can be handled easily with their respective package managers.

Snap

#

Simply open the Snap store to install software effortlessly, though the packages are often outdated.

apt

#

# Software source operations can be performed graphically in the desktop version via "Software & Updates"
# Add a software source
sudo add-apt-repository ppa:libreoffice/ppa && sudo apt update

# Remove a software source
sudo add-apt-repository --remove ppa:libreoffice/ppa

# Install software  
sudo apt install xxx  

# Update packages  
sudo apt update # Sync package info from remote repositories without upgrading  
apt list --upgradable # View upgradable packages  
# Upgrade all available packages without handling dependency changes  
sudo apt upgrade  
sudo apt full-upgrade # Full upgrade  
sudo do-release-upgrade # Upgrade across major Ubuntu versions  

# Check software packages
sudo apt-cache search wps # Search for all packages containing "wps" and their description information
sudo apt-cache pkgnames | grep -i wps # View package names containing the keyword "wps"

# Remove packages  
sudo apt remove xxx  
sudo apt autoremove # Clean up residuals  

If there is an abnormality in network access, it might be necessary to manually specify the proxy settings for apt. By default, apt does not use the system’s proxy settings; instead, you need to explicitly write the proxy settings into the /etc/apt/apt.conf.d/99proxy file.

# Check the system's proxy settings
env | grep -i proxy

# Open the configuration file
sudo nano /etc/apt/apt.conf.d/99proxy

# Write the proxy settings, filling them out according to your system's proxy settings
Acquire::http::Proxy "http://127.0.0.1:7890/";
Acquire::https::Proxy "http://127.0.0.1:7890/";

deb

#

After downloading a .deb package from a browser, double-clicking it will install it directly. Internally, this uses apt, so management is the same as with apt.

# Install via double-click  

# Uninstall via apt  
sudo apt remove xxx  
sudo apt autoremove # Clean up residuals  

AppImage

#

AppImage is a portable software packaging format for Linux systems, designed to simplify application distribution and execution. Its core philosophy is “one app = one file,” allowing users to run applications directly without installation or administrator privileges.

In newer versions of Ubuntu, attempting to run the file directly may result in an error. To resolve this, you need to install libfuse2 using the following command:

sudo apt install libfuse2

Then grant executable permissions to the AppImage package:

chmod +x file_name.AppImage

After installation, you can simply double-click the package to launch the application 😃.

If you want to uninstall the software, it’s very straightforward: just delete the package. However, if you’re a perfectionist like me, you can check the following directories to completely clean up any residual files:

ls ~/.config -a # Check configuration files

ls ~/.local/share -a # Check shared configuration files

ls ~/.cache -a # Check cache
du -sh ~/.cache/* | sort -h -r # Check disk usage of folders under .cache

curl

#

Download and execute installation scripts directly from a URL using curl. Software installed this way is harder to manage because the actual installation process is script-driven and difficult to monitor.

# Example: Installing the Zed editor  
curl -f https://zed.dev/install.sh | sh  

# Uninstalling is usually messy  
# First, fetch the installation script  
curl -f https://zed.dev/install.sh -o install.sh  

# Have an AI parse the script  
# Then follow the AI’s instructions to manually uninstall  

Major Version Updates

#

Performing major version updates is completely unnecessary for server OSes, as the related software packages usually haven’t caught up yet. Chasing the “latest version” isn’t wise. However, for desktop users, it’s quite useful—after all, updating allows them to experience the newest system features. In short, it’s just for fun. 🤓

This section primarily references the WeChat public article: How to Upgrade from Ubuntu 24.04 to Ubuntu 25.04.

Data Backup

#

This step is essential. Although it might take up dozens of gigabytes of disk space, a major version update is still a risky operation. Better safe than sorry. 😅 You can always delete the backup and free up space after a successful upgrade.

# Install the backup tool  
sudo apt install deja-dup  

# Run directly  
deja-dup  

Update Software Packages

#

Ensuring the system is up to date minimizes compatibility issues. Execute the following commands one by one:

sudo apt update  
sudo apt full-upgrade  
sudo apt autoremove  
sudo apt autoclean  

sudo reboot # Reboot the system  

Version Upgrade

#

The logic is straightforward: point the old version’s software sources to those of the new version. Below are some relevant files that may need modification:

• Upgrade policy file: /etc/update-manager/release-upgrades
• Software sources configuration file: /etc/apt/sources.list.d/ubuntu.sources

If you want to upgrade from an LTS version to a non-LTS version, you’ll need to modify the policy file. The policy file actually contains just one line—change it to the following:

Prompt=normal  

Next, modify the software source configuration file by running the following command:

sudo sed -i 's/noble/oracular/g' /etc/apt/sources.list.d/ubuntu.sources  

After modifying the files:

# Refresh the index and perform a full upgrade, including the kernel, drivers, and all packages  
sudo apt update && sudo apt full-upgrade -y  

After the upgrade completes:

sudo reboot # Reboot to apply changes  

lsb_release -a # Verify the system version  

Office Suite

#

As we all know, Microsoft Office cannot run directly on Linux 😅. However, viewing and editing doc files is often unavoidable.

Therefore, here’s a recommended Office alternative for Linux: LibreOffice. The installation steps are as follows:

sudo add-apt-repository ppa:libreoffice/ppa  
sudo apt update  
sudo apt install libreoffice  

Before installing LibreOffice, I also tried using WPS to edit Office files, but for some reason, it kept causing system errors, so I eventually abandoned it.

Storage Cleanup

#

Common Cleanup Tasks

#

# Remove orphaned dependencies  
sudo apt autoremove  

# Clear apt cache  
sudo du -sh /var/cache/apt # Check apt cache size  
sudo apt autoclean # Auto-clean  
sudo apt clean # Full clean  

# Clear system logs  
journalctl --disk-usage # Check system log size  
sudo journalctl --vacuum-time=3 d # Remove logs older than 3 days  

# Clear .cache
du -sh ~/.cache/* | sort -h -r # Check cache size
rm -r folder_name # delete the folder recursively

# Clean up old Snap versions
snap list --all # List all Snap packages

# List all disabled packages (single-line command)
echo -e "\033[1 mDisabled Snap Packages and Their Sizes:\033[0 m" && snap list --all | awk '/disabled|已禁用/{print $1}' | while read -r pkg; do size=$(snap info "$pkg" | awk '/installed:/ {print $4}'); printf "%-30 s %10 s\n" "$pkg" "$size"; done | sort -k 2 -h

# Remove all disabled Snap packages (single-line command)  
snap list --all | awk '/disabled|已禁用/{print $1, $3}' | while read snapname revision; do sudo snap remove "$snapname" --revision="$revision"; done

# Clean up old kernels  
sudo dpkg --list | grep linux-image # List all kernels  
sudo apt autoremove --purge # Automatically remove unnecessary kernels  

Auto Clean on Boot

#

Manually checking and cleaning up your system every time can be a hassle. A smart computer should learn to clean itself 😋

Run the following command to open a new script file:

sudo nano /usr/local/bin/system-clean-up.sh

Then paste the following content into the file:

#!/bin/bash
set -e

echo "[1] Running apt autoremove..."
apt autoremove -y

echo "[2] Running apt autoclean..."
apt autoclean -y

echo "[3] Cleaning journal logs older than 2 days..."
journalctl --vacuum-time=2d

echo "[4] Removing disabled snap revisions..."
snap list --all | awk '/disabled|已禁用/ {print $1, $3}' | while read snapname revision; do
  echo "Removing snap: $snapname revision $revision"
  snap remove "$snapname" --revision="$revision"
done

echo "[5] Cleaning ~/.cache/ directories larger than 200MB..."
for userdir in /home/*; do
  cache_root="$userdir/.cache"
  [ -d "$cache_root" ] || continue
  for dir in "$cache_root"/*; do
    if [ -d "$dir" ]; then
      size_kb=$(du -s "$dir" | awk '{print $1}')
      if [ "$size_kb" -gt 204800 ]; then
        echo "Removing large cache directory: $dir ($(($size_kb / 1024)) MB)"
        rm -rf "$dir"
      fi
    fi
  done
done

echo "[Done] Clean-up finished."

This script includes five safe automatic cleanup tasks:

1. apt autoremove
2. apt autoclean
3. Deletes system logs older than 2 days
4. Removes disabled Snap packages
5. Deletes .cache subfolders larger than 200MB

After saving the file, grant execute permission:

sudo chmod +x /usr/local/bin/system-clean-up.sh

Then configure automatic startup on boot: The directory /etc/systemd/system contains service scripts that run automatically at startup, all with the .service extension.

To enable the script to run automatically at system boot, create a new systemd service file:

sudo nano /etc/systemd/system/clean-up.service

Paste in the following content:

[Unit]
Description=Clean up system caches, logs, and snaps at boot
After=network.target

[Service]
Type=oneshot
ExecStart=/usr/local/bin/system-clean-up.sh
User=root

[Install]
WantedBy=multi-user.target

After editing is completed, run the following command to enable this service script, so it will automatically run once upon the next boot but will not run immediately.

sudo systemctl daemon-reexec && sudo systemctl enable clean-up.service

You can also run it manually to test if it works properly:

# Run immediately
sudo systemctl start clean-up.service

# Check the system journal
sudo journalctl -u clean-up.service

One-Click Auto Setup Script

#

If you find the above script configuration process too cumbersome, we also provide a one-click automated setup script. You can delete it after running it once.

Create a file named setup-cleanup.sh in any directory and paste the following content:

#!/bin/bash

set -e

echo "🚀 Creating cleanup script at /usr/local/bin/system-clean-up.sh..."
cat << 'EOF' | sudo tee /usr/local/bin/system-clean-up.sh > /dev/null
#!/bin/bash
set -e

echo "[1] Running apt autoremove..."
apt autoremove -y

echo "[2] Running apt autoclean..."
apt autoclean -y

echo "[3] Cleaning journal logs older than 2 days..."
journalctl --vacuum-time=2d

echo "[4] Removing disabled snap revisions..."
snap list --all | awk '/disabled|已禁用/ {print $1, $3}' | while read snapname revision; do
  echo "Removing snap: $snapname revision $revision"
  snap remove "$snapname" --revision="$revision"
done

echo "[5] Cleaning ~/.cache/ directories larger than 200MB..."
for userdir in /home/*; do
  cache_root="$userdir/.cache"
  [ -d "$cache_root" ] || continue
  for dir in "$cache_root"/*; do
    if [ -d "$dir" ]; then
      size_kb=$(du -s "$dir" | awk '{print $1}')
      if [ "$size_kb" -gt 204800 ]; then
        echo "Removing large cache directory: $dir ($(($size_kb / 1024)) MB)"
        rm -rf "$dir"
      fi
    fi
  done
done

echo "[Done] Clean-up finished."
EOF

sudo chmod +x /usr/local/bin/system-clean-up.sh

echo "✅ Cleanup script created."

echo "🚀 Creating systemd service clean-up.service..."
cat << EOF | sudo tee /etc/systemd/system/clean-up.service > /dev/null
[Unit]
Description=Clean up system caches, logs, and snaps at boot
After=network.target

[Service]
Type=oneshot
ExecStart=/usr/local/bin/system-clean-up.sh
User=root

[Install]
WantedBy=multi-user.target
EOF

echo "✅ Service file created."

echo "🔄 Reloading systemd..."
sudo systemctl daemon-reexec
echo "✅ systemd reloaded."

echo "🧩 Enabling clean-up.service at boot..."
sudo systemctl enable clean-up.service
echo "✅ Enabled."

echo "⚙️ Running cleanup task now..."
sudo systemctl start clean-up.service

echo "✅ Cleanup complete. You can check logs with: sudo journalctl -u clean-up.service"

Finally, run this setup script once and you’re done:

sudo chmod +x setup-cleanup.sh && sudo ./setup-cleanup.sh

Install Nodejs

#

Node.js is a commonly used environment dependency and often takes up a lot of disk space. If you don’t understand its installation logic, it’s easy to leave behind a large number of useless files.

My attitude towards such basic environment dependencies has always been: if you don’t understand it, never install it.

The standard installation command can be found on the official website:

# Download and install nvm:
curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.40.3/install.sh | bash

# Reload nvm in the current terminal
. "$HOME/.nvm/nvm.sh"

# Download and install Node.js:
nvm install 22

# Check Node.js version:
node -v # Should be "v22.16.0".
nvm current # Should be "v22.16.0".

# Check npm version:
npm -v # Should be "10.9.2".

# Optional, execute only if pnpm needs to be installed👇
# Download and install pnpm:
corepack enable pnpm

# Check pnpm version:
pnpm -v

The first command uses curl to download the nvm manager, which is a tool for managing Node.js versions. This tool will be downloaded to the ~/.nvm directory. Specific usage methods will be introduced later.

The second command is used to load nvm in the current command line so that the nvm command can be effectively recognized. The reload operation can also be done without a command by simply closing and reopening the command-line window.

The third command uses nvm to install Node.js. The main node packages will be placed in the ~/.nvm/versions/node/ directory, for example, the 22.16.0 version will be placed in the ~/.nvm/versions/node/v22.16.0/ directory.

When using nvm to install node, the corresponding version of npm will be installed by default, so there is no need to install npm separately 😊

pnpm requires an additional command to install, as shown above. A brief introduction to pnpm is provided below: Introduction to pnpm

Introduction to nvm

#

Similar to other package management tools, the use of nvm itself is not complicated, and its purpose is to stably and faithfully complete the most basic management work. Below is the most common command table:

If there are any other special requirements, you can check them on nvm’s official GitHub: nvm

Introduction to npm

#

npm is a package management tool for Node.js, used to manage node dependency packages in projects.

In actual project development, we use npm to install some dependencies, which can take up a lot of disk space. Below are the main cleanup items:

##### First item, node_modules #####
# Run in the corresponding root directory to view space usage
du -sh node_modules

# Cleanup method is simple (clean as needed)
rm -r node_modules

##### Second item, global dependencies #####
# Query the location of global dependencies
npm root -g

# View the size of global usage
du -sh $(npm root -g)

# Cleanup method is simple (clean as needed)
rm -r module_name

##### Third item, download cache #####
# View cache location
npm config get cache

# View cache size
du -sh $(npm config get cache)

# Clear cache
npm cache clean --force

Introduction to pnpm

#

pnpm is another Node.js package manager, similar to npm and yarn, but it is more efficient in terms of performance, disk space usage, and dependency management.

Common command table, including comparison with npm commands:

Compared to npm, pnpm manages space more cleanly. Below are some key directories:

Since pnpm does not store actual files in node_modules, there is no distinction between global and local caches. A single command is all you need to clean up all removable dependencies:

# Clean the cache
pnpm store prune

pnpm scans all .pnpm-lock.yaml files in the projects you’ve used, identifies unused versions, and cleans up the corresponding cache. If needed, this command can be directly added to Auto Clean on Boot.

Miscellaneous

#

This section includes some simple yet commonly used commands.

System Control

#

• Shut down immediately: shutdown now or systemctl poweroff

• Restart immediately: sudo reboot or systemctl reboot

Generally speaking, commands using systemctl are more “graceful,” allowing running processes to end autonomously before shutdown, rather than being forcibly terminated.

Here are two recommended aliases for shutdown and reboot:

alias seeu='systemctl poweroff'
alias again='systemctl reboot'

Extracting Files

#

The command varies depending on the file format you need to extract.

# Extract a .zip file  
unzip file.zip -d /target/directory  

# Extract a .tar file  
tar -xvf file.tar  

# Extract a .tar.gz file  
tar -xzvf file.tar.gz  

Temporary Environment Variables

#

Sometimes, even with a VPN proxy enabled, the terminal may still be unable to access the internet properly. This is likely due to the terminal being unable to automatically detect the proxy port. The solution is to modify the environment variables, which can be changed permanently (by writing to a file) or temporarily, as shown below:

# For use in the PowerShell terminal on Windows
$env:HTTPS_PROXY="http://127.0.0.1:7890"

# For use in bash or cmd terminals
set HTTPS_PROXY=http://127.0.0.1:7890

Temporary File Inspection

#

For regular files, you can just open them with the corresponding editor. However, for some special files, such as extremely large CSV files, you may only want to quickly check a small portion of the data. In this case, you can use the head command:

# Read and display the first 10 lines of your_file (without reading the entire file)
head -n 10 your_file.csv

If garbled text appears, it must be due to an encoding format issue. You can use iconv for conversion:

# Convert from GBK to UTF-8
head -n 10 "your_file.csv" | iconv -f gbk -t utf-8

If you’re unsure about the original and target encodings, just throw the garbled text to an AI — it’s simple and convenient 😋

References

#

• How to Partition and Mount Disks in Ubuntu
• LibreOffice Suite
• 在 Ubuntu 安装配置 Fcitx 5 中文输入法
• Configuring Git to Push by Default Without Entering Credentials (Ubuntu, SSH)