Reviving a netbook (Part I)

Part II Part III/I

For quite some time, I've toyed with the idea of giving my very ancient netbook, an eMachines eM250, a second life as, what I like to call it, a “tiny typewriter”. Due to its pleasant keyboard and the fact that is much more lightweight than any of my proper laptops, it still may fulfill a decent job as a simple typing machine I can take with me during field trips without having to worry too much about accidentally breaking it.

About the machine

While this netbook was used for various stuff back in 2011, not only did it ship with Windows 7 Starter, a very crippled variant of the original 32-bit OS, but its charger port became loose and got pushed rather deeply into the case, preventing the original charger from reaching the port. Attempting to repair it turned out to be impossible due to the plastic case slowly beginning to break and certain sections appearing to be glued together, not allowing me to push a knife between said case and the upper section with its keyboard. A full disassembly – removing every piece bit by bit – still is an option, yet after noticing that I cannot simply separate the top from the rest like I can on my Asus laptop, I quickly became frustrated and, as a last resort, I grabbed my Acer's charging cable with its bigger plug, kept it in place against a table leg, and, even to my own dumb surprise, I managed to get this old machine fully charged after a decade of not using it with a full battery.

The issues, however, run much deeper than this silly hardware fault. In some parts of the world, this device originally shipped with Windows XP, highlighting its much lower requirements than its successor. Even Windows 7 Starter is way too heavy for this machine, with Starter consuming a whopping 646 MB of RAM out of 1 GB when idling. CPU consumption jumps between 21 and 34% when running the file manager and the fan gets relatively loud. Wifi was disabled during the latest check but there are little performance differences between enabled internet connectivity and full offline usage. Overall, I don't know how I managed to tolerate this for roughly two years, considering I – as a pre-teen – already knew that Windows 7 Starter doesn't allow even the most fundamental user settings such as changing the background.

(It should be noted, though, that I was that kind of young, inexperienced user that downloaded every web browser toolbar in existence, never checked any downloads from BearShare, didn't understand that effective backups are NOT stored on separate users on the same machine, and demonstrated a rather questionable taste in music and aesthetic preferences, namenly “Blingy-fied” pictures. At least I managed to rcover my old ICQ number.)

Preparations

Before jumping straight into the real deal, I first had to properly archive some files on my incompatible, external HDD. Noticing that the vast majority of files are duplicates (or... let's say, some files exist at least five times) but happen to be scattered around various folders and even users, the backup procedure already would require some time. So before I took care of this, I decided to plan a testing routine, giving various Linux distributions suited for the old 32-bit architecture a chance. As Puppy Linux already was tested a few years ago and did not fully satisfy me, it did not make it into the pre-selection. MX, despite supporting i686, is not included, either. The main reason for this is its 64-bit variant recently having seen a sudden increase in system requirements with its latest release – considering the age and hardware limitations, MX, at this time, may be too heavy.

The current list of preferred distributions includes:

Among those that may be considered in case the preferred selection is not sufficient:

You may have noticed that both lists exclude any distribution relying only on systemd. While I do worry about its feature-creep whilst being dependent on it on all of my 64-bit machines, my primary concern rather deals with the possibility of future incompatibilities with its 64-bit variant and current system requirements. Keep in mind that this netbook can barely run Windows 7 Starter and its weak battery, so systemd, which is heavier than any other init and init/service manager combination, may claim precious resources needlessly. Recently also having taken interest in Gentoo and OpenRC, I've never given any other init besides the speedy s6 (via Artix) and SysV (via Slackware) a run at least within a virtual environment. Suffice to say, I'm only really familiar with systemd and I would like to try something different.

Despite 32-bit systems now only ever being encountered among embedded computing, the choice of available distributions still is fairly solid. If I were to include systemd-based distributions, this list would be longer and include classics such as Linux Mint, Debian, openSUSE and Mageia. Distrowatch lists quite a lot of 32-bit distributions for i686-based machines (which, if the naming scheme, may confuse you easily, can run any OS supporting processors between the very ancient i386 and the last generation, i686) that still are being maintained today, so coming up with a strict list of requirements was necessary.

As I already am familiar with some of the preferred distribution, those will be granted higher priority (antiX, Devuan, Slackware). As Parabola is based on Arch Linux, which currently is my main operating system, it will be tested first, as a successful test would make system maintenance across all of the machines I use much easier due to sharing the same package manager.

VirtualBox setup and test strategy

For most of my previous tests, I intentionally went with an unrealistically-underpowered standard configuration for each virtual machine to check each distributions minimum requirements. Even among pure 64-bit operating systems, this (admittedly ridiculous) configuration proved to demonstrate that a careful Desktop choice already makes even a midweight such as Debian quite usable on weak hardware. All VM's are assigned a single CPU, core and ~1.5 GB of RAM, the maximum VirtualBox will allow to allocate on my Medion tower. Only tests done on my relative's HP Pavilion slightly exceeded this standard limit.

As this low VM setup is almost identical to my netbook's specs, all upcoming tests may perfectly emulate the following “bare-metal experience”, so greater focus will be put on the virtualized tests to weed out potentially unfitting candidates and settle with one or two distributions for the bare-metal tests.

The distribution satisfying me the most will remain on this netbook and further customized to turn this machine into my desired “tiny typewriter”.

If everything fails

Despite high optimism, I will not discount the possibility that two things may happen during those tests or afterwards:

a) No distribution fits my criteria or it will take too much effort (more than a day) to set it up, b) the loose charging port gets pushed even deeper into the machine and my dirty workaround will no longer work.

While I don't think that the first will happen, given how stable the Debian ecosystem is and how the number of users running Alpine as a Desktop appears to be growing, the latter may happen at some point, either now or years after this. Only then I'll go don the full-disassembly route and save the parts that still can be used, while the rest goes straight into the recycle bin. Obviously, due to parts of the case being sealed I likely will got avoid to break it apart, which will make re-assembly more tedious and time-consuming than it should be. As much as I'd like to give this netbook a second chance, this machine isn't worth spending months (and aching hands) on it, either. Its HDD still is functional and my ancient 32-bit tower still got plenty of slots left to host a second hard drive, which then could be used for further distribution tests.

(The only downside to the alternative plan is the tower running on an ancient NVIDIA GPU and its issues caused by the last driver's update, which made me abandon this machine as a kid. The latest drivers caused the graphics' card to shut itself off after five minutes and the only way to solve this was by downgrading the driver. Windows Vista, however, kept pushing updates re-installing the faulty driver, yet my local Aldi sold a cheap Multimedia tower plus monitor, keyboard and mouse at the same time, which I ended up receiving and still use today. I needed a stable machine for school work back then and this 32-bit tower just didn't do it not only in terms of software stability. Now that Vista has long been EOL and the machine's no longer dependent on an internet connection, it's easy for me to avoid the faulty driver permanently under Vista, though I'm not sure how Linux distributions will perform. It surely is worth an investigation at some point.)

Part II will deal with the actual tests.