for a while now i've had an old public computer i stole from my old job (call it severance for trying to garnish my wages), however only recently have i managed to use it for anything useful, other than running my partner and i's minecraft server. that was of course before microsoft stole my minecraft account, but that's a different article.
lately i've been spending my free time actually developing it however, and i've realized how much i've slept on the helpfulness of docker. years ago as a child running an online radio for my discord server, an icecast2+mpd deployment powered by a beheaded laptop on my home ip, i remember it being a total nightmare (for a twelve year old) managing dependencies, keeping track of my system architecture, how different services interact, and how to upkeep it beyond letting it sit. after finally learning docker, and then promptly docker-compose, i have such a newfound appreciation for containers and their applications. the ability to almost instantly summon a fully functional service from rotely copying a yaml file from github is stunning to me. furthermore the ability to immediately erase a service the moment i grow tired of it with no residual schmutz left on my device is wonderful. i've never been particularly good at cleaning house. docker is a delight for me, and has truly enabled me to finish projects that have sat on the backburner for nearly half of my life. of course this came with the hurdles of learning docker, but for such a widely used and praised piece of software it's remarkably simple and understandable. the most difficult part is properly understanding what all the options and separations between host and container entail. of course, the year currently being 2026, i can't say i ruggedly rtfm and just figured it out. the temptations of the beast call to us all, and in this part, i've succumbed wholeheartedly.
my vice of poison being deepseek, it is shocking how much usefulness ai actually has once you deflate the hype and the ongoing reconstruction of babel. whereas in prior years i would've spent likely multiple evenings troubleshooting, pouring over stackoverflow, and trawling through project forums, deepseek allows me to simply enter in the problem i'm facing, some context details as necessary, and a couple of error logs and it gives me the simple answers to the dumb problems that would normally have me pulling hair out. of course you could argue, rightly i might add, that by offloading some capacity of understanding and troubleshooting to ai i am thus degenerating my own capabilities to perform, and robbing myself of the sweet nectar of knowledge that comes with pulling teeth until your qbittorent mod properly sets the forwarded port. to this i say you're completely right, and i have very medium shame. on the contrary, if i were to propose an excuse i would argue that internet search is dead and the giants of indexing have openly admitted to hiding quality results on later pages to keep you on the site longer. it's genuinely shocking the uselessness of google anymore, i never thought i would see the day that duckduckgo could be truthfully argued to provide better results. besides the mere obfuscation of information, ai has the patience that no foss irc chat has ever had with me, in that i can spam error messages back at it and it will endlessly give me alternate solutions rather than tell me to build it from source. this of course isn't to mindlessly fellate llms and glorify them as a whole, i am widely critical of the current ai bubble, its propaganda, and its greater detriment to the whole of society. But if ai can reasonably troubleshoot stupid problems for my applications, then i'm sure it can explain to the nice old lady that she can't connect to the printer because her bluetooth is off, and if the only positive ai has in society is that it takes a load off the backs of those who installed minecraft mods as children, then perhaps it's all worth it.
what initially spurred this interest on was fucking off at work on lobste.rs and discovering hister. hister bills itself as a personal search engine that indexes your private knowledge-base. a nightmare for a closed source product is once again a delight as a foss project. truthfully my main complaint with the surveillance state isn't that they track me everywhere i go and build an in depth personalized model on my searches, purchasing history, locations, and whatever else can be sold to advertisers, it's that i dont have access to any of the fun statistics or data. i am thus always happy to track myself and what i do. while i have not had a proper chance to use hister for this purpose yet, i am happy knowing that should any of the eclectic websites i've found go offline tomorrow, i have a full snapshot of the site the last time i visited it. i'll never lose a flash game again. i enjoy much of hister's design currently, however i do have a couple sticking points so far. hister has great tooling built for modern web browsers, that's about where the tooling ends. beyond downloading the extension or importing your history, you unfortunately have very little ability to add data in other ways. there is of course the index command that, due to the beast we spoke of earlier, immediately bounces off every website you attempt to crawl, however i find such a focus on websites and browsers to be lackluster of something which describes itself as a "personal search engine." the file importing seems very lackluster, although i have not personally used it very in depth, and it supports a meager group of services i have no interest in using, much less snapshotting. hister is a new project, and while i may be disappointed with it's current nubility, i am very excited to watch the direction it's heading towards the future. after finishing my work on hister, i found the nice thing about docker being so extremely approachable and easy to use, is that once you successfully finish one project, i can't help but feel unsatisfied until i start another.
and so i decided to revive attempts to get a private jellyfin server operational so i can watch everything they think they can make me pay paramount for. and once more jellyfin was extremely easy to get working with docker, requiring very little fine tuning or adjustment. the main issue i discovered however was that i quickly tired of transferring media over the wifi with scp (yes i'm aware of rsync, no i don't know how to use it yet). thus enters my white whale; a seedbox to torrent media without optimum threatening to shoot my dog. as long as i've known about plex servers i've longed to have my own service where i can seamlessly queue up media for download and have it instantly appear ready to play on my tv without finding an hdmi cord and remembering how to use xrandr. i've stumbled over my own feet numerous times in the past attempting to get a setup working for me, even purchasing a protonvpn subscription for it, but i had never managed to figure out the tooling for it and how to mesh it neatly with a headless device. i may have a national award in networking infrastructure, but i know nothing about it, and i can only hope the rest of the contestants didn't either. yet again however a foss project came to my rescue with a yaml config i could rotely copy and a container to do all the hard work for me. after finally learning that the README.md was outdated and gluetun had changed their authentication processes, i managed to get the entire stack working. qbittorent with its traffic ran through gluetun running all traffic through a wireguard connection to protonvpn servers, with jellyfin watching its download directory ready to add media. of course i could, and would like to, go farther, i've been reading on the *arr services and have some interest in them, although my main issue is i don't plan to download everything on earth (all of this was free after all, i haven't even bought hard drives), but the ability to search for media and have it automatically download without hunting for the correct version i want on whatever tracker will give it to me, does sound pretty nice. my partner has been deriving much joy from this as well, using jellyfin makes one remember the years before streaming services attempted to stop being loss-leaders and were simply good value for your money. a complete lack of ads or interruption to service while also being able to deliver superb quality with no buffering is an experience that is hard to beat, much less for the price of free.
i'm unsure of where to go exactly from here, i've scrolled through the millenial awesome selfhosted lists and haven't seen anything particularly interesting at the moment, although they are extremely incapable of properly advertising software. i'd like a couple more network based services and perhaps home automation things, but those will have to wait until i have more fine-grained control over my networking situation. working on my home server has genuinely been such a delight for me lately, and has greatly improved my local mental health. for a long while now i've been struggling to create any meaningful art, much less any i'm proud of, and it's very rewarding to have something that i can both feel is mine, and feel some pride over. it's even rejuvenated my will to live enough to finally write another article for this site, which i too am proud of. i would promise to update the site further and perhaps try and figure out a solution for hosting my photography, but i know better than trusting my whims.