[Day 02] - My Journey from Laptop to Lab -- Server setup successfully..

This is the second day of my Journey make my laptop server, If you are not read the first version yet kindly read that article first [Day 01] - My Journey from Laptop to Lab: Building a Headless Debian Server for DevOps & AI

Alright now it’s time for try the new OS, In day 01 you might remember that I downloaded and update the bootable device with Debian 13, then let’s see what to do today.

I start the install debian-13.3.0-amd64-netinst and when I go through the process, I got steps like setup domain, setup admin password (/root) and setup user account by giving the full name, username, password.

Then I came the process to setup the disk, At that process the Linux installation ask me to whether follow the guided setups or manual steps to install the system. I selected the manual method, because wanna fresh installation.

Then I got the following screen,

The current disk status

I have the deleted all the partitions in both SSD (sda) and the HDD (sdb), and I created two manual partitions for the server to run,

536MB - EFI System partition
127.5GB - ext4 partition for server

The EFI System Partition (ESP) is a critical, OS-independent partition that serves as the storage location for bootloaders, kernel images, and drivers used by the system firmware during startup.

As I found about the ESP, this act like a bridge between the firmware and the operating system. instead of a "boot sector," the ESP holds the actual .efi binary files (like grubx64.efi) that the motherboard loads to start OS.

This is also a platform independent and I found that a single ESP can hold bootloaders for multiple operating systems (like Windows and Linux) if needed.

Refer the image how that looks after the server install successfully. (This is after server setup - you are time travelling).

But there is a something notable, that I haven’t setup the swap for my installation, When I did little search about the swap, I discovered that I can either setup the swap when I install the OS or after setup the OS.

So, I decided to find what are the difference between these two approaches and what will be the suitable for me.

Here what I did found,

Setting Up Swap in Installation (Partition Method)

During the installer, we would create a dedicated "Swap Partition" on SSD, that method associated with following advantages and disadvantages.

Advantages	Disadvantages
Performance: A dedicated partition can be slightly faster than a file because the data is stored in contiguous blocks on the disk.	Inflexible: Once create a 2GB partition, it is difficult to resize it later without unmounting drives and using tools like GParted.
Reliability: It is "set and forget." The system will always have that emergency memory available from the first boot.	Wasted Space: If we never use swap, that 2GB–4GB of fast 120GB SSD is permanently "locked" and can't be used for AI models or files.

Setting Up Swap Later (Swap File Method)

Once Debian is installed, we able to create a "Swap File" (a large file sitting on root partition that acts like RAM). This method associated with following advantages and disadvantages,

Advantages	Disadvantages
Dynamic Flexibility: This is the modern "DevOps" way. We can create a 1GB swap file today, and if we find Llama-3.2 needs more, we can delete it and create an 8GB file in seconds without rebooting.	Slight Overhead: Theoretically, a file has a tiny bit more overhead than a raw partition, though on modern Linux kernels and SSDs, this difference is practically zero.
No Disk Fragmentation: We can put the swap file on your SSD for speed or move it to your 1TB HDD if we want to save SSD space.	Manual Setup: We have to run 4-5 terminal commands after the installation is finished.

If we compare the overall process in both approach it looks like this,

Feature	Setting Up Now (Partition)	Setting Up Later (File)
Ease of Setup	Automatic (via Installer)	Manual (via Terminal)
Flexibility	Low (Hard to resize)	High (Easy to resize/move)
Disk Usage	Fixed "slice" of the disk	Just another file on the system
Performance	Slightly better	Practically identical on SSDs

To came to the conclusion, I got selected the Swap file method, since my Dell has 8GB of RAM, I likely won't even need swap for a minimal Debian server and small AI models like Llama-3.2 1B. There for I skipping it now.

Then If I hit a "Memory Out" error later while running a models or working I plan to create a swap file later.

Before getting a decision I also consider about the performance difference in the both approaches, Here what I got through Gemini,

Metric	Swap Partition (Set up now)	Swap File (Set up later)
I/O Speed	Maximum (Direct access)	Near-Maximum (Minimal overhead)
Access Latency	Consistent	Negligible difference on SSD
System Boot Time	No impact	No impact
CPU Overhead	Zero	Extremely low

For me “Swap File” method seems its flexibility for future DevOps projects as well, because I can adjust the swap storage as needed in future.

Then I continue and I follow the few more steps like select the mirror server and other minor settings.

Finally I got to select the software that want to install to the OS, In that screen I only selected the SSH Server And Standard Utility to keep the my server more lightweight but make sure essential software are installed.

SSH server is essential for managing server from remotely and standard system utility included the essential command line tools like following,

Key Categories of Included Tools

Text Editors: Basic terminal-based editors like nano or vim-tiny for editing configuration files.
Networking Tools: Utilities for managing connections and troubleshooting, such as iputils-ping, curl, and wget.
File Management: Tools for handling archives and file systems, including tar, gzip, bzip2, and parted.
Process Management: Basic monitors like top to view system resource usage and manage running tasks.
System Documentation: Access to manual pages through the man-db system to look up command usage offline.
Hardware Detection: Small utilities that help the kernel identify and interact with your specific laptop components.

That’s all after few minutes I got my brand new (yet laptop is old) server, with lot of memory to play for me. In simple term, now I have bigger ground to play, look at my resource usage now 👇

What do you think ? It’s glad to hear yours opinion on the comment section 👇

Alright that’s it for today, let’s meet in day 3 with more exciting story, most probably my next goal is to setup llama, let’s see what happen next. see you all in the day 03.

Thank you for reading… ! 😎