ryukamish

Switching to a Free-ier ISP

Thu, 13 Aug 2026 00:00:00 +0000

Carrier-grade NAT are also known as CGNAT or large-scale NAT (LSN) which are used by the two major ISPs here. It’s great for normie internet consumption but is an ass when trying to host services. As there’s no single IP address assigned to a single user. This makes self hosting on the internet just non-existential. More accurately saying, CGNAT provides IPv6 addresses only.

There’s also an article on dev.to website where they explain how they host their homelab and expose it to the internet through IPv6.

Raspberry Pi as the server

There are many problems with not using the official Linux images which Raspberry suggests. I had variable success of installing Debian, OpenBSD, FreeBSD and Alpine. Sometimes it worked sometimes it didn’t. It could have been a skill issue but installing Debian should be easy and as stress-free as possible. For me it was just a rainbow colored bootloader screen. I also did update the firmware through the official Raspberry Pi image.

Alpine has worked more times than other distros. Debian and FreeBSD didn’t go through the rainbow bootloader screen. As of now my Raspberry Pi is running Alpine Linux. It’s lightweight and is just 350MiB on a new install.

UPDATE: I’ve switched to Debian for Pi and laptop. This was majorly because of more documentation available and tested by users. Also, I’m lazy and don’t have enough time to diagnose and fix issues on Alpine. I fixed the installation issue by using the official Raspberry Pi Imager to flash SD card.

Dynamic IP Address

My ISP doesn’t give a static IP but a dynamic one. It’s another charge on top of the internet which I already pay for. It’s mostly a waste of money until there is something very serious to host which requires 100% uptime. Services like a commercial website where you sell things, payment systems, etc. I created three virtual servers on my router. One, which listens on port 80, second which listens on port 443 and third which listens on port 22.

I fixed it with ddclient. It runs in the background as a service and automatically updates the IP address at a fixed interval. It was the most simple thing to install and configure. Although, be careful of where you edit the config file. There are two, one in /etc and another in /etc/ddclient. The file inside directory ddclient is the one you want to edit. I’ve wasted so much time diagnosing a problem which never existed.

There’s a YouTube video of Bugswriter where he showcases how he hosts his personal website and his business website with Google Cloud Provider. In his video he was using his own Rust program to update his IP address for his websites.

Web server and SSL certification

The very first thing before hosting a website is to install a web server which listens on HTTP and HTTPS ports. HTTP requests are directed to port 80 and HTTPS requests to 443. HTTPS is HTTP protocol over TLS/SSL. For my use case, I only want to expose the HTTPS port for my website.

Firstly, install nginx and certbot. After nginx is listening on port 443 and service is reachable in the local network, use certbot --nginx -d website.com to generate SSL certificates for website.com. Certbot requires a working domain on the internet. If only local SSL certification is required mkcert is an option which could be used. Many web hosting providers offer HTTPS with their internal tools. In this case, Certbot is required as I’m hosting my website on my own home server.

Cloudflare DNS

I’m using Cloudflare as my domain registrar and DNS provider. It’s mostly because of convenience rather than choice. I just have to fill in A records with my hostname and IP address. I have proxied my website, which servers my website through Cloudflare hiding my IP address.

Also, check out Cloudflare DNS records page for knowing what DNS Records are and the types which are most commonly used.

Cost of hosting

The internet plans are relatively cheap compared to other ISPs. Also, they have more variety in internet plans. The power bill is low overall as the server is a Raspberry Pi which draws ~2.9 W in idle. It’s manageable.

The cost of domain name is rather expensive compared to the power bill. I bought it for $1.5 for a year. It will be ~$2 for a year after this. It doesn’t look much in terms of dollar but the conversion rates are rather high for me.

Overall, the cost of hosting my website is cheap compared to buying a domain with hosting. One could get started with hosting their own website for very cheap. It’s a one time investment for a Pi which I bought second hand for $4.5. If it’s something like my ISP provider, getting a router is free. Just the recurring cost of internet.

Git Server Setup

Thu, 23 Jul 2026 01:30:06 +0530

Setting a Git instance on the local network is relatively easy compared to setting up cgit. I was on it for hours with more failed attempts than success. Following the git book to get git on a server is easy. It’s just copy and paste a couple of commands. cgit is a little different.

Setting a Git server

The best way to setup a Git server is by following the official handbook. Still, to quickly setup a Git server make a bare git repo with git init --bare <repo>. It creates a bare repo (this is important) which shouldn’t have a .git directory.

Firstly, create a new user with a different home directory. It’s better to have a new user rather than use the same one because of security reasons. Also, change the owner to of the git directory to the user created.

# On Alpine Linux
adduser <user> -h <home dir>
# On systemd based distros
useradd -b <home dir> <user>

If only a local git repo is required then this is enough. Just add the public key generated through ssh-keygen of local machine to the server’s ~/.ssh/authorized_keys directory. After this when pushing to the server’s repo password won’t be required.

Setting up cgit

I’m using Alpine Linux for my Git server so many commands are specific for it. Install cgit with the package manager with additional packages. spawn-fcgi and fcgiwrap are also required. The official Alpine wiki provides a better and extensive details on setting cgit.

Installation

# On Alpine
apk add cgit git fcgiwrap spawn-fcgi

After installing these packages edit /etc/cgitrc to add some very specific lines. These include the root directory where it will read from.

virtual-root=/
scan-path=/path/to/git/repositories/

scan-path value is the directory where cgit will read from. Giving the root directory where all the git repositories are present is a better options rather than individually adding them.

Also, add a service file in /etc/conf.d called spawn-fcgi.cgit. It runs fcgiwrap automatically everytime it’s called. It should exactly look like the lines below.

FCGI_PORT=1234
FCGI_PROGRAM=/usr/bin/fcgiwrap

Then, start the service with rc-service spawn-fcgi.cgit start. It automatically runs everytime the service is called. Even after rebooting it runs the service again.

Running the web service with nginx

I’m on nginx for my web server. There are other options like lighttpd, apache, etc. For me nginx was what most guides provided and other people also use it so documentation and people’s experience with it generally gives better troubleshooting support.

The Alpine wiki suggests adding a file in /etc/nginx/conf.d but I’ll be using the /etc/nginx/sites-available directory for this service. This is the way I’m most familiar with. After creating that file in sites-available symlink it with ln -s /etc/nginx/sites-available/git.your_domain.com /etc/nginx/sites-enabled/. Now it’s serving on port 80 of the server’s ip.

server {
listen 80;
listen [::]:80;
server_name git.your_domain.com;
root /usr/share/webapps/cgit;
try_files $uri @cgit;
location @cgit {
include fastcgi_params;
fastcgi_pass localhost:1234;
fastcgi_param SCRIPT_FILENAME $document_root/cgit.cgi;
fastcgi_param PATH_INFO $uri;
fastcgi_param QUERY_STRING $args;
}
}

Restart the nginx server with rc-service nginx restart.

Now, just add domain name git.your_domain.com to where you manage your domains to point to localhost on port 80. It should bring up the domain on the internet which is running the web interface for the git server.

Miscellaneous options

To add more functionality to the web interface other options could also be added. It follows the same convocation of option=value. Edit the /etc/cgitrc file to add these functionalities.

project-list - Only show the specified git repo mentioned here. Add the full path file name.
remove-suffix - Removes the .git at the end of the directory name.
enable-http-clone - set the value to 1 to enable cloning through http link.
agefile - set it to info/web/last-modified. This requires a post-receive hook in hooks directory of the bare repo.

Contents of post-receive hook in hooks directory.

#!/bin/sh
mkir -p info/web
date '+%Y-%m-%d %H:%M:%S' > info/web/last-modified

Using ProtonVPN in Sway

Fri, 26 Jun 2026 14:05:00 +0530

I have been using ProtonVPN for a year now. I have also used it on Windows. The better used operating systems i.e. Windows and MacOS have a decent app which lets you connect to different countries. The app experience on Linux is also great if on a Desktop Environment like KDE Plasma or GNOME. I also used it on Hyprland and Niri and it worked as expected. But, on Sway everything changes from just works to make it work. It’s more work to get Sway working.

It matters how it’s launched. If proper variables are not given there can be some programs which will not launch. Also, launching with dbus-run-session gives a lot of problems. I was launching it with dbus and gnome-keyring autostart didn’t worked properly. It would always prompt me for a password for a default keyring. Launching it without dbus fixes the issue. I fixed the issue with running it without dbus.

Wireguard is the better way

Using wireguard is the better way of connecting to a vpn. I say so because it doesn’t require a dependency on a keyring and most of the times it just works. Ironically, GUI apps have OOBE but here in the Linux world a terminal command is the better way of doing the same thing as far my opinions are concerned.

Installing wireguard

On Arch, wireguard can be installed with pacman -S wireguard-tools. The Arch wiki provides more details, tips and tricks and troubleshooting guides.

As for Gentoo , it can be installed with emerge --ask net-vpn/wireguard-tools The Gentoo wiki has more info on compiling flags, USE flags and other troubleshooting guides. Include +wg-quick USE flag to compile with the command to control the connection.

Getting conf file from VPN provider

Get the conf file from the VPN provider which contains the private key and public key which will allow a secure connection between the server and the client. It also contains the DNS, Address, Endpoint, etc.

Put the conf files in /etc/wireguard directory with preferred name. Make the names easier to type as the name without the extension name will be used to connect through wireguard.

Going under the tunnel

Now, use the command wg-quick up <name> to connect to the wireguard instance. It will require root privileges to use this command.

Conclusion

Using a GUI app on systems with no desktop environment and is DIY like Sway one could avoid all the headache of dependency, environment variables and auto start programs. I’ve spent hours on solving the keyring issue for ProtonVPN with gnome-keyring but something or the other fails. As for my VPN connections I only use one out of 4 to 5 VPNs at a time. So having options is good but not needed for my use case.

Another thing to consider is the use of dnscrypt-proxy and dnsmasq with wireguard. Most of the times doesn’t work. I got it working for already visited webpages but not able to connect to a search engine or any other webpage which I haven’t visited already.

Gentoo Linux as a daily use machine

Thu, 25 Jun 2026 15:55:58 +0530

Note: This is an opinion piece and everything written here is specific to me. Experiences may vary for others. Some things could be true to me but not necessarily to others.

A while ago I just made the switch from Arch to Gentoo. It took me 3-4 days to get to a functional system with graphical interface.

Getting to a tty only took a day which was quick tbh. I followed the Gentoo wiki with another guide from マリウス Gentoo installation . He used the hardened version of Gentoo in his install guide.

USE flags

One of the many powerful features of Gentoo is USE flags. They include or exclude what the binary of a program has. Like if someone is not using desktop environments KDE or GNOME one can exclude them from the build with USE flags. Another example is excluding systemd from a OpenRC init system. I primarly switched to Gentoo because of USE flags.

There is this annoyance when using separate USE flags for every individual package. This creates manual maintenance for packages’ USE flags and dependencies. I’ve let go of this as a new user to Gentoo. It was too much work to see the dependencies and manage the USE flags for different versions of the packages.

I went with the --autounmask-write and dispatch-conf for managing packages. This is the least headache and at the end of the day a functional system is what I want not following an idle blindly.

Package Manager

The package manager is written in Python. It’s generally slow to run but the compile times take longer than the language to execute the file in /usr/bin. It’s a single file of ~100 lines.

I’ve generally found out that searching for packages is slow with emerge. Say, for example, if you want to search for gentoo-kernel package. It takes around ~1-2 seconds to look for the package in the repository and give back some output. One second or two isn’t that bad to be honest but when trying to search for packages continuously it becomes a hassle.

I’m also not saying that other package managers are fast or faster than emerge but it becomes a rhetorical question of what package manager is the best.

Managing audio with OpenRC

This one is tricky as systemd used to manage audio and bluetooth in Arch Linux. Gentoo provides a script called gentoo-pipewire-launcher which should be started when graphical display starts. The best way when using Sway is to launch it with a script which will have all the variables.

Also, there was an update to pipewire which added a new sound-server USE flag which could be found on the Gentoo website or through eselect news read. It enables/disables the sound capabilities of Pipewire. It was done mostly to resolve the conflict between Pulseaudio and Pipewire. There was an unintentional migration from Pipewire to Pulseaudio and vice-versa. Basically, to resolve the taking control over the audio without the user knowing.

As the Pipewire dependency is growing with time needed for crucial functionality like screen casting, etc.

Building packages

Let’s say, for example, there is a package which you want to install and it’s not available in the Gentoo and GURU overlay. Also, it has something to do with the kernel like adding a module for you laptop. Building these packages can be a pain not because it’s difficult but to make sure it works as expected dependencies have to be resolved.

In my case, I was installing a module for my MSI laptop but adding modules to the kernel requires recompiling it with those kernel flags. I’ve been using the binary version of the kernel which most probably has the support built in but a source compiled kernel will need to be recompiled.

I’ve also built packages from source which are not in Gentoo repos. They didn’t seem to have any issues but other applications which only provide the binary version can have issues. Some applications require cups to be installed. The application doesn’t have anything to do with printing services but still it requires the package to be installed.

Upgrading packages

Upgrading Gentoo is a little different from other distros like Arch, Fedora, Debian, etc. The very first thing to do is sync the remote repository with the local repository. This could be done with emaint.

emaint --auto sync or emaint -a sync can be used to sync with the remote repository.

After syncing the repository upgrade the packages with emerge --ask --verbose --update --deep --newuse @world or emerge -avuDN @world.

When upgrading any configuration changes can be addressed with dispatch-conf command. It’s the easy way to manage the USE flags for individual packages. Also, --backtrack=N can also be used to when there is dependency issues with upgrading packages. It can be used with --emptytree which will take more time to upgrade packages.

A full official Gentoo page has all the options which contains other sections for other options.

TL;DR

Using Gentoo as a daily driver can be a pain when you don’t have much time to spare to compile software, diagnose package compatibility and resolving conflicts. Gentoo as other DIY distro is time consuming and you’re responsible for your system. If something works or breaks it’s because of you. I’m back on a binary based distro Arch. There were times when I wanted convenience of just installing a binary. And, yes, Gentoo does have binary but it also requires resolving package conflicts.

Personally, for me I don’t have a deep desire to only use a source based distro in order to reduce the binary size to absolute necessity. I can live with having bigger binary sizes with features not in use. Maybe, I can use it on my server to make the distro very niche to that system. ChromeOS based on Gentoo (Archive)

Neovim

Mon, 15 Jun 2026 10:35:29 +0530

Defaults are powerful. Before using any type of plugin learning the defaults without any plugins is important as not every time personal configs are available on the system and knowing how to navigate around the editor with the defaults is very important.

Defaults

Neovim without plugins is also usable. The default file manager Netrw is also usable. Although, for many advanced and convenient features and workflow plugins come in handy. Another handy editor commands are opening a split pane in horizontal and vertical split view with :vsp and :sp for horizontal and vertical split.

Convenient key motions

In addition to the usual hjkl motion other motions are also present. Moving in Neovim with style to quickly move in the buffer and edit changes.

Using ciw to change a word and go into insert mode. Helpful when changing a word and cursor is in-between text
Using vi" to select text inside quotes.
Using t to move forward to the character before the cursor.
Using f to move forward on the character after the cursor.
Using T to move backward to the character after the cursor.
Using F to move backward to the character before the cursor.
While in visual mode one can use o to go between the selected line.

g key motions

This motion is so awesome. Suppose there is big text on the same line and I want to go to the beginning of the visual line, I could go like g0. This goes to beginning of the visual line. SO COOL! Another example, is when I want to go to the end of the visual line. I can use s$ to go to the end of the visual line.

Another example of this is when moving between lines. I can use gj to go down exactly below the cursor. Generally, using j and k moves very differently, but this fixes that.

Another great example is using gq to format a big line into sentences in newlines. This is very helpful in files like markdown and any other text based file types.

ftplugin directory

There’s a builtin spell check which can check grammar in English and other languages which I have not tested. Spellcheck for all types of files doesn’t make sense. Spellcheck can used in normal files like a text file, markdown which are more verbose in grammar.

For only enabling it in markdown files, create a file named markdown.lua in after/ftplugin with

vim.opt_local.spell = true
vim.opt_local.spelllang = { "en_us" }

content in to it. This file will only enable spellcheck in markdown type file. The opt_local is used to make the settings applicable only to the current markdown filetype.

There are also keymaps which can be used to correct the wrong spellings with

]s to jump forward
[s to jump backward
z= suggests correct spellings
zg to mark the word as good
zw to mark the word as bad

To target plain text files one can use text.lua file which targets plain text files. I have setup wrap, linebreak and spell check only for plain text files.

TL;DR

I’ve learnt a lot from TJ DeVries from his Advent of Neovim YouTube videos. I have also learnt a lot from ThePrimeagen’s Neovim YouTube videos which are available on YT for free. There is also the official documentation which also has a lot info on how to use Neovim.

Setting Up DNS

Thu, 04 Jun 2026 00:00:00 +0000

Note: I’ve found the hard way of using dnscrypt-proxy with with ProtonVPN is not a good idea. ProtonVPN changes resolv.conf which has the IP address for NetworkManager. While on the other hand this guide locks the file itself making it prone to change. For the time being I don’t know how to do this. Although, using dnscrypt-proxy without locking resolv.conf file works with ProtonVPN.

The first things to do is install the packages. If on Arch Linux install them first. I’m not using docker because it will be awkward to host both. Port 53 binding is annoying in docker as privileged ports from a container adds friction and on systemd/OpenRC systems systemd-resolved or another stub listener often already holds port 53.

DNSCrypt setup

Installing packages in Arch

Install both packages with sudo pacman -S dnscrypt-proxy dnsmasq. This will install both packages.

FYI: dnscrypt-proxy runs locally as a DNS resolver, encrypts DNS queries to make sure that no one connected to the same domain see what domains I look up. It supports DNS over HTTPS (DoH) and DNSCrypt protocols. dnsmasq on the other hand is a DNS caching to gain speed on the local network. It caches the encrypted DNS queries so repeated lookups are answered quickly.

Setup DNSCrypt config

Now editing the config file in /etc/dnscrypt-proxy/dnscrypt-proxy.toml file. Add the lines inside the file or edit the changes.

# Change listen port from 53 to 5300 (avoid conflict with dnsmasq)
listen_addresses = ['127.0.0.1:5300']

# Pick resolvers — these are no-log, DNSSEC-validating
server_names = ['cloudflare', 'quad9']

# Recommended privacy options
require_dnssec = true
require_nolog = true
require_nofilter = true

The port generally is bind to 53 but as we are also using dnsmasq port should be something different. dnsmasq reads queries from port 5300.

dnsmasq configuration

The general file to edit is in /etc/dnsmasq.conf.

# Don't read /etc/resolv.conf
no-resolv
# Forward all queries to dnscrypt-proxy
server=127.0.0.1#5300
# Listen only on loopback
# Also adding waydroid interface
interface=lo,waydroid0
listen-address=127.0.0.1,192.168.240.1
bind-interfaces
# Cache size (number of entries)
cache-size=1000
# To enable negative caching
neg-ttl=60

When the user makes a query it is forwarded to port 5300 which then searches the internet and brings back results which are then cached in dnsmasq as entries.

Stop anything holding port 53

On CachyOS port 53 is generally bound to systemd-resolved. Check with

ss -tulpn | grep ':53'

If systemd-resolved is present then run

sudo systemctl stop systemd-resolved
sudo systemctl disable systemd-resolved

Fix /etc/resolv.conf

systemd-resolved manages this file as a symlink. Replace it with

sudo unlink /etc/resolv.conf
echo "nameserver 127.0.0.1" | sudo tee /etc/resolv.conf

Then lock the file so that NetworkManager cannot overwrite it.

sudo chattr +i /etc/resolv.conf

(For if any reason you want to restore the file then run sudo chattr -i /etc/resolv.conf)

Enable the services and verify

Enable both service via systemctl with

sudo systemctl enable --now dnscrypt-proxy
sudo systemctl enable --now dnsmasq

Verify if it works

# Check both are listening on the right ports
ss -tulpn | grep -E ':53|:5300'

# Test resolution through the full chain
dig google.com @127.0.0.1

# Check dnscrypt-proxy is actually encrypting (should show your resolver)
sudo journalctl -u dnscrypt-proxy -n 20

In the journal you should see lines like [cloudflare] OK (DoH) or [quad9] OK (DNSCrypt) confirming encryption is active.

When using NetworkManager, it may try to overwrite the resolv.conf and try to manager DNS. To prevent that one can edit /etc/NetworkManager/NetworkManager.conf and add

[main]
dns=none

Then restart NetworkManager with sudo systemctl restart NetworkManager

Undervolting on MSI Modern 14 C13M

Tue, 02 Jun 2026 10:44:59 +0530

Warning: Anything which requires to tweak any hardware level settings can be damaging to the computer. This can void warranty (if any) and can seriously brick the system which could lead to not booting up. There have also been cases of firmware damage. Do it at your own risk!

Internals of MSI Modern 14 C13M

This laptop emits a lot of heat for how less things I do with it. On idle it sits around 50°C. If there is time where I need to compile something the temperatures go above and beyond 80°C, which is wild. Although, I understand the temp going up but for a laptop it shouldn’t go this high.

I wanted to control the heat from this laptop which lead to undervolting. I researched it well when I was doing it but only after a while did I got to understand that undervolting the CPU itself was not possible w/o bricking the system. Overclock is the other end of the spectrum which boosts the clock base speed. This in return gives maximum performance and a lot of heat. Overclocking reduces the life span of the CPU.

Controlling power limits

One effective way to control heat is to limit the power used by a computer. A low powered machine doesn’t require much energy. I tried running the below commands and for a moment everything was working normally until apps started freezing and opening apps took more time.

To check what are the numbers already set by the kernel run the below commands.

# Main package (usually the one that matters)
cat /sys/class/powercap/intel-rapl:0/constraint_0_power_limit_uw # long-term (PL1)
cat /sys/class/powercap/intel-rapl:0/constraint_1_power_limit_uw # short-term (PL2 burst)
cat /sys/class/powercap/intel-rapl:0/constraint_0_time_window_us

# Optional: energy used so far (for monitoring)
cat /sys/class/powercap/intel-rapl:0/energy_uj

# See live package power (install linux-tools if missing)
sudo pacman -S linux-tools
sudo turbostat --show PKG_Watts --interval 2 --num_iterations 5

To temporarily check if the modified values took place or not one can run below commands.

# Moderate – good balance (recommended starting point for Modern 14)
echo 28000000 | sudo tee /sys/class/powercap/intel-rapl:0/constraint_0_power_limit_uw
echo 35000000 | sudo tee /sys/class/powercap/intel-rapl:0/constraint_1_power_limit_uw # short burst a bit higher

# More aggressive cooling / battery focus (very quiet, still usable)
echo 22000000 | sudo tee /sys/class/powercap/intel-rapl:0/constraint_0_power_limit_uw
echo 28000000 | sudo tee /sys/class/powercap/intel-rapl:0/constraint_1_power_limit_uw

# Ultra conservative (max battery, very low heat – good for office/web)
echo 18000000 | sudo tee /sys/class/powercap/intel-rapl:0/constraint_0_power_limit_uw
echo 22000000 | sudo tee /sys/class/powercap/intel-rapl:0/constraint_1_power_limit_uw

All these manually running and hoping the values are correct for the system will not do. This could lead to problems. Machines can do a lot of things on their own nowadays so manually intervening seems kind of dumb.

My recommendation is using tools like auto-cpufreq to manage CPU frequency and power draw. It automatically changes the frequency on certain or set conditions set by the user.

Tools used for undervolting

auto-cpufreq
turbostat - included in Linux kernel

Tl;dr

It didn't go as planned. Undervolting on my MSI laptop is not possible. It is locked and cannot be unlocked without bricking the system. I’m using power profiles daemon for power management.

Experience on Cachyos

Mon, 01 Jun 2026 23:23:49 +0530

It’s been a year of using CachyOS on my main laptop as a daily driver. Never faced any major issue which cannot be fixed. The system is snappy and quick to respond. The effects of customized packages and custom kernels builds are quite visible. Although base Arch in itself is very light and snappy. For a normal user who doesn’t play games or compile software will not see much difference in speed between Arch and CachyOS but is still worth while to have those optimizations.

Firewall enabled by default

Unlike Omarchy which has ufw installed and rules written but it doesn’t enable it by default. The user needs to enable it manually. That’s already a plus for security.

GUI for almost everything

A GUI application is installed called Shelly which installs, updates and manages packages in CachyOS. It’s a GTK app. For me, searching of packages were made easier than using a terminal.

Pre-configured system

CachyOS already provides a lot of options to use as a desktop in Calamares installer. Many popular options like GNOME, KDE Plasma with Window managers like hyprland , niri , etc. These are already configured for ease of use. The window manager are also configured with mostly sane defaults.

Bookmarks

Mon, 01 Jun 2026 19:46:55 +0530

Tech

linux.die.net - Linux documentation

Videos

Websites

マリウス - Personal website of Marius
terminaltrove.com - A huge collection of TUI apps
Edward Snowden Substack
classics.mit.edu - Read Classics Literature

Sleeping Late Causes Problems

Fri, 29 May 2026 12:27:00 +0530

Why does it cause problems?

The main reason is the lack of sleep reduces the receptors of our brain which makes us slow. Reading a paragraph doesn’t make sense until you read it again and again. But, for some reason on other days you could understand those lines pretty quickly.

This is because our body repairs the damaged cells during a good night sleep which were damaged during the day.

Circadian Rhythm

This is the internal clock of the body. It controls the wake and sleep cycle. It tells your body when to sleep and when it’s time to wake up. It also controls hormones, digestion and body temperature. It’s like have a personal butler inside your body who checks whether everything is going smoothly. When you don’t have a regular sleep and wake cycle then hormones are disrupted like melatonin which is a sleep hormone.

Mood and Motivation

It also affects mood causing mood swings. One moment you’re angry another depressed. It also drains away the willpower to do things.

About

Mon, 01 Jan 0001 00:00:00 +0000

Hosted code

I host my code on Github and Codeberg. I also have my own Cgit instance.

ryukamish

Switching to a Free-ier ISP

Raspberry Pi as the server

Dynamic IP Address

Web server and SSL certification

Cloudflare DNS

Cost of hosting

Git Server Setup

Setting a Git server

Setting up cgit

Installation

Running the web service with nginx

Miscellaneous options

Using ProtonVPN in Sway

Wireguard is the better way

Installing wireguard

Getting conf file from VPN provider

Going under the tunnel

Conclusion

Gentoo Linux as a daily use machine

USE flags

Package Manager

Managing audio with OpenRC

Building packages

Upgrading packages

TL;DR

Neovim

Defaults

Convenient key motions

g key motions

ftplugin directory

TL;DR

Setting Up DNS

DNSCrypt setup

Installing packages in Arch

Setup DNSCrypt config

dnsmasq configuration

Stop anything holding port 53

Fix /etc/resolv.conf

Enable the services and verify

Verify if it works

Undervolting on MSI Modern 14 C13M

Controlling power limits

Tools used for undervolting

Tl;dr

Experience on Cachyos

Firewall enabled by default

GUI for almost everything

Pre-configured system

Links

Bookmarks

Tech

Videos

Websites

Sleeping Late Causes Problems

Why does it cause problems?

Circadian Rhythm

Mood and Motivation

Links

About

Hosted code