<----HOME

FreedomTest History

This section explains many details about how the concept of FreedomTest was conceived and the beginnings of the server, but also expands into more recent events, alternating between my point of view as administrator and the point of view of players. This section will be updated around once a year.

Origins of the server

During 2019 I was learning about web technologies; mostly HTML and CSS languages. At the time I wanted to setup my own website to get more experience in web development, but also to learn about server administration, so I dug up an old Pentium 4 PC that was collecting dust at home and I decided to turn it into a reliable server for hosting my website. Since Pentium 4 processors output a lot of heat it needed to have enough cooling, hence I bought fans for the rear and front of the case, with the latter also cooling the hard drive. Mounting the front fan required drilling mounting holes into the case because it was not designed for using front fans. Another requirement was to make it resilient to power outages since those are pretty common where I live, so I decided to use solid state storage for the drive that will keep the OS. I could have used an SSD but the budget was limited and instead I went for a Kingston 32 GB USB 3.0 flash drive which was enough for my goal of serving a small static website, where the actual contents would be in a RAM drive. The picture shows the computer with the case fully disassembled for testing the fans before mounting everything else. This computer features a 3.06 GHz Prescott Pentium 4, 1280 MB of DDR RAM, a Foxconn motherboard with a P4M800 VIA chipset, a chinese refurbished power supply and later on a 500 GB Seagate hard drive was added. The server was initially tested and assembled in October of 2019, but due to lack of time to dedicate to the project I had to delay designing the website and setting up the web server. Instead I decided to use it as a NAS and installed the Seagate drive in it, this happened in the first week of 2020 and it was when I installed Debian Buster in the Kingston drive. After that the original project was being delayed again and again because of lack of time for it. Now it is when Minetest gets into play, I had already come across Minetest in 2019 but I did not like that performance was bad on my computer so I forgot about it, however during the pandemic in 2020 around March/April, in times of boredom I rediscovered it and decided to give it another try, and this time I tinkered more with it and started to like it more. Also it was during this time that I came across 2b2t videos on YouTube and I got engaged with those videos. At this time the server was already abandoned, doing basically nothing, and I thought it was a perfect fit to use it to host a Minetest anarchy server, this is how it all started. During the next months I was just testing configurations for Minetest, as well as mods and learning about how to make a Minetest server work. My goal was to get the server online as soon as possible to take advantage of the spike in popularity of anarchy servers on YouTube, but at the time I only had access to my home ISP which is not good enough for hosting a Minetest server so I had to wait to be able to deploy the server with another ISP.

First days of FreedomTest

Time was running out and I had been busy with other things so I decided to launch the server using my home ISP. I got everything together during a weekend in September of 2020 and brought the server online for the first time on September 14th around 04:30 GMT. I tried some world seeds before ending up with the current one because I thought that floatlands was not working, in one of those seeds spawn was an island in the middle of the ocean; that would have made escaping spawn an interesting challenge. On the first day 4 players joined the server and it also crashed because of some bug related to the engine or the airtanks mod. After that many players visited the server but most of them only stayed for 10 minutes at the most. By November I had already tested the server enough to be confident on its performance, so I decided to announce it on the Minetest Forums and the Minetest Reddit subforum on November 21st, though shortly after the announcement a power outage occurred and people were not able to join. Thankfully two weeks later I got some help and was able to deploy the server in a different city with a faster ISP and a UPS that enables the server to gracefully poweroff and hibernate when a power outage occurs, assuring no data loss. During the rest of the year the server got many more players thanks to the public announcements and it seemed to be going fine. Back when I started the server for the first time I came up with the name FreedomTest quickly without giving it much thought, the word Freedom was chosen because the server is meant to be a place where players can do whatever they want and Test just because of Minetest.

2021

Even though the announcements helped getting more people to know the server, not many players were staying and it was stalling, so I decided to contact Minetest Videos in January 2021 for him to make a video showcasing the server. He ended up making a short series of videos that really brought a ton of players into the server, and the Pentium 4 machine along with the USB boot drive were starting to show their limitations so I started looking for an upgrade. During February 2021 the server reached the highest amount of player activity so far. At the time the server was running with a better quality power supply that I borrowed because the original was problematic, causing unexpected shutdowns, but I had to return it and replaced it with the original one, which unfortunately failed around mid February, and since I did not have physical access to the server there was nothing I could do to fix it, so I had to set up a temporary server running on my desktop computer while I got the chance to travel and fix the server. Since the Pentium 4 server was already being overwhelmed and had failed I took the components from my desktop computer, namely the motherboard which I had already decided was going to be the hardware upgrade and replaced the old motherboard with the newer one. The server kept the same case but now it had an AsRock motherboard with an Intel 945 chipset, an E7200 dual core CPU and 4 GB of RAM out of which only 3.3 GB are usable due to chipset limitations, however the upgrade was not flawless because the BIOS of this motherboard is not as good as the one from the Pentium 4 motherboard and it cannot reliably hibernate when using a USB drive as boot drive, so I had to disable hibernation in a hurry to get the server running, then in early March the server was up and running again with the original world. Unfortunately things were going to get worse, as around mid March the upgraded server failed, again I did not have access to it and was not able to fix it, also I did not have much time available to dedicate to this project and the server stayed down for almost a week. Because of the COVID-19 pandemic travel was restricted and I was not able to leave my place, so I set up a backup server at home using the Pentium 4 motherboard and some other old components with which I got the server online again on March 18th, running a temporary world. By this time most regular players had left because the server was down for too long. This temporary setup was in place for almost two months, and when I was finally able to recover the failed machine along with the world files, the original world came back on April 30th, running on the Pentium 4 server at home which now was playing the role of backup server. Later I discovered the cause of the server failure was that the AsRock motherboard was refurbished with new old stock capacitors manufactured in 2004 which were already degraded, causing the motherboard to fail. I will continue updating this section later this year from my point of view. Following is a diary provided by the player MooseMan which describes many events spanning from 2020 to the first failure of server hardware:

Server started in November of 2020 on a Pentium 4, single core 2.8 GHz CPU with 1.2 GB of RAM. It runs on some form of Linux, completely without desktop on Debian Buster. The server's connection was really poor, and over 50% of the time wasn't accessible. Shortly after christmas of 2020, the server had a couple internet connection upgrades, and now the server runs significantly better. The admin, Minix, says that the server will be upgraded at some point to a core duo 2 core at 2.16 GHz with 4 GB of RAM. On December 27th, 2020, for the first time 4 people were online at once. The server was still on original hardware and was pretty crappy. When this happened, damage, eating, doors and saturation all stopped working and had a minutes long delay. Boats were unusable, but regular walking was unaffected. Blockbreaking wasn't too bad. The people online were: MeguminZ, MooseMan, Pizzaria, and Sanity. On January 2nd 2021, 5 people were online for a brief period of time. The server had the upgraded internet connection and it runs really well compared to the first time 4 people were online. MooseMan, MeguminZ, Pizzarria, KmanLucas, and C12plus. The internet connection works pretty well and I can't wait to see the upgrade of this server, and for it to get utilized more. On January 17th 2021, three major events happened. 1. The members of Frosty Village griefed spawn, while the youtuber "Minetest Videos" filmed and enjoyed our chaos. 1066 TNT blew up directly around 0,0 and about 50-60 lava buckets were dumped around spawn. 2. At one point, 7 people were online, the members of frosty village, NathanS21, Beerholder, and a couple others. 3. The lava was so intense, and with 6 people on the server at once, loading multiple chunks in different places, the server crashed, and once we reconnected, it crashed again. (The server's fine a few hours later). After emailing the admin freedomtest@protonmail.com, I found out that during our event, the server's CPU was between 80 and 100% all the time for about an hour. The two crashes were not actually caused by us lagging the server, but by the travelnet mods, which I noticed weird behaviour with. When MeguminZ places the travel net box by the spawn I could not enter it until he came to the frosty village.

The admin says the minetest world is a cube of 64000x64000x64000 assuming that each chunk (20 million blocks) takes up 5MB, there is potential for 655 TB of space. Unfortunately, most of this is air, so there isn't anything noteworthy over there. Assuming all ground level chunks are loaded, it's only 10GB of disk space. The way to make higher disk space without loading any new chunks, is by writing in books, or by making complex builds that are hard to compress. Assuming the mechanical hard drive is 128GB of space and that there is a 1:10 compression ratio, assuming ASCII encoding where 1 character takes up 1 byte. We'd have to have a fucking long book to fill up the hard drive. There may be a hardcoded limit to these books, say maybe 64K characters in length. Now let's talk about holes. Digging to the bottom of the world with our current quarry machines, assuming a 1x32000x1 hole, would take 8.8 hours. We could fill this hole up with TNT to add a significant amount of randomness to make it harder to compress, or we could get a large number of quarry machines and run them all at once. As 64x32000x64 hole would take 4096 quarry machines 8.8 hours to dig. The fastest, but most intense way to dig the hole would be by hand, being AFK guarding the quarry machines would likely be easier and faster. Moose was digging a hole at 3103, 3687 on the 18th of January 2021. About 4K down, shortly after loading the Slade biome, there was a bad chunk that crashed the server every time he logged in. He crashed the server over 5 times in one day. This means two things: There could be random bad chunks all over the server. Or the df_caverns mod may load some bad chunks. It could be an isolated event however. Shortly after this event, I broke through the slade layer, (Digging with TNT) and I decided to use up the last of my supply, and once it went off, the server crashed. It was the same issue however, I loaded a bad chunck, and I contacted the admin. There was an issue with the df_caverns plugin, and after he updated it, there were no more issues with bad mapblocks. As of January 22, 2021, the hole is about 24K blocks deep. On January 24th 2021, the server was down for about two hours. It turns out, this was to upgrade the hardware. The new server has two RAID1 hard disk drives, a Pentium Dual Core E5500 (2.8GHz), and 3GB of RAM. Playing later that day, we noticed that the server was very laggy, and often would disconnect us, or refuse to connect in the first place. As of noon AST the day after, the internet connection seems to be stronger, albeit still significantly more unstable than the old server. It is the author's humble opinion that if it is a desktop, the network adapter should be swapped so a more stable experience can take place. I am optimistic about the server, and can not wait to see what the future brings. RTT (Round Trip Time) = 408ms I believe the old server was 40-60. On January 26th, this problem was solved in large part. At first the admin thought that it was the backup ISP that was causing the issue. He switched to the backup ISP because the first one ran out of bandwidth (20GB/month) It turns out that it wasn't the root cause, and it was an issue with the router. There have been a lot fewer connection time outs this afternoon because of this fix. I would also like to make a side note. Between the 17th and 29th of January, a lavacast was made around 0,0 by the player Deathwalker. On the 31st of January, MooseMan used an alt account to take away all the lava. As of the 31st of January, the world map file was 1.7GB large, an increase of 400MB in a single week. I expect the mapfile to be over 3GB by the end of Febuary. On the 1st of Febuary, 2021 the admin Minix came online to reduce damage from lava and mese by 40% because too many new players were dying because of the lava around spawn. On the 5th of Febuary 2021, was the first time the slade layer was broken through by solving the puzzle. It requires flowers in the right order. The colours matter and there is a colour code hidden in the source code of the df_caverns mod. As of the 6th of Febuary, 2021 the world map's file size was 2.1GB The increase likely because of the influx of new players and moose exploring the slade layer heavily. The admin says he's also keeping an eye on a nether plugin and that says that he'll include it when he updates the server's version of minetest to 5.4.0. It is unknown when this will happen.

A very concerning conversation took place between the admin and the author over the weekend of 6th and 7th of Febuary 2021. Due to the rising popularity of the server, the bandwidth limits of the server are rapidly approaching. Last month the admin transferred the worldfile to a second computer on a second ISP in a different city, but the compression times took 6 hours. This was at 1.7GB, and by the end of the month assuming a worldfile increase of 300MB a week would take over 10 hours. Unfortunately, the bandwidth won't last until the end of the month. As of yesterday, 11GB was used out of the 20GB, and linear interpolation says that the data will run out early next week. Luckily the admin says that once the server's data usage hits 15GB, he will start to upgrade to 40GB/month. His current plan costs 10USD/month, and to cover the increasing costs of running the server, the admin says he will open a donation page. Assuming 1% of players will donate and the average donation is 5USD, it will take 400 regular players to completely cover the costs. However, it's likely that 50 regular players will likely cover most of the costs of the server. The admin also wants to reward the players that do donate, but isn't sure how to reward them. The author suggested giving them system information like RAM/CPU usage, (most important to me, HDD usage) The second reward suggested by the author, is the administrative anchor block for the technic mod, which would save bandwidth, but increase CPU usage. This would be convienent for some players, but the admin would have to carefully moniter CPU usage, and may lead to conflict with players. Update at 10:00PM EST the author noticed that the maximum players allowed online was reduced from 10 down to 4. On the 9th of Febuary 2021 the admin imposed a cerfew on the server and it is only active from 17:00 to 5:00 UTC. The server was upgraded to 5.4 for a couple days over the weekend of the 12th. But it was reverted back to 5.3.0. The admin said on the 20th of Febuary that within a week he will upgrade to 5.4.0 and also add some new mods. The world map file is 2.5GB, a smaller increase over the past two weeks due to the lower player count since the cerfew was imposed. The admin and the author are both optimistic on the future of the server. But the unthinkable happened. The server went dead. Radio silent. Febuary 25th the server went completely down for over a day, and the day after, I logged in, but it prompted me to create a new account. Then I logged in, finally, a completely new map was generated. I have taken several pictures, if you want to see them, hit me up and we can work something out. The admin used this map to test a couple new mods, and used it to upgrade to 5.4.0. Afterwards the admin made a vote that he left up for 36 hours.