> Sonarr, Lidarr, Radarr, etc. Every time I read about setting it up, it just sounds like a resource-intensive mess that’s fundamentally prone to frequent breakage. That’s the opposite of what I’m looking for when choosing self-hosted services.
I just checked on my *arr stack. Sonarr and Radarr idle at 157MB and 159MB of memory and approximately no CPU. Jellyfin currently idles at 330MB. I'm not sure where you got this impression from. I grant these tools are complicated and have somewhat clumsy UX, but they aren't especially heavy to run and are not fragile.
OP, point your :robot: at https://github.com/subdavis/selfhosted/blob/main/docker-comp... and ask for an explanation. It's a single compose file that illustrates how the *arrs, transmission, and jellyfin all work together. I'm sure you can set it up :)
Indeed. The other thing to point out is that, once you have set them up, they will chug along fine for years with near zero input.
Here is the stack I recommend:
* Radarr (movies) with optional second instance for 4K versions
* Sonarr (shows) with optional second instance for 4K versions
* Sonarr (anime), separate instance to have control over quality settings
* Prowlarr (tracker/indexer management)
* SABnzbd (Usenet)
* qBittorrent (torrents)
I would also recommend consulting the TRaSH guide to tune your setup and ensure you’re grabbing quality releases. From there, the next logical step is Profilarr :)
“Even better” is relative. Some people like having media stored locally instead of relying on torrent streaming availability.
Usenet streaming sounds cool, but with releases being DMCAd so frequently, I don’t see how it would be practical for anything but the most recent content.
Also, I imagine that Stremio without Debrid (or equivalent) is kind of a terrible experience most of the time.
Yep basically. The real magic is when you combine it with a Debrid service, which essentially puts a cache in front of torrents. If the torrent has been downloaded before, you get a direct hosted stream instead of having to stream your movie from active seeders (i.e. buffering hell).
The interfaces of arr stack are ugly, and linking qbit to indexer manager to arr app is needlessly complex. Maybe there could be plugins in one app like qbit.
I used to think so too, but now I usually just have Muse or another AI agent on my box do those things for me. I should maybe create a slimmed down API first *arr stack equivalent then hmm
You can configure quality profile via the tags too (I do have separate profiles for anime vs liveaction).
I didn't even set the tag, as I don't intermingle them in the same directory, and automatic tagging by sub path is possible too
So I'm also surprised why someone would configure a second instance for that
If using the TRaSH guides (synced automatically with recyclarr), with all their quality profiles etc, I find that those mostly don't make sense for anime. It's easier for me to keep the anime specific stuff separate, different naming conventions, different episode numbering, etc.
There's been two major disruptions in my experience.
First was Lidarr went for months last year (was it this year?) unable to search/add musicians due to complications with how they were relaying those searches. It took them a long time to refactor that part of the software and make it usable again.
Second was the retirement of Readarr, the ebook-version of this software, which was an absolute shame.
But aside from that this is pretty much set-and-forget software. Although the setting is harder than it should be, should be much easier to favour smallest x265 rips at <resolution>, lossless audio, season rips with subtitles.
In this day of vibeslop, it's insane no one thinks to combine all these into single app. The best we can do is having single docker compose file that has to manually point each service into each other.
A single dream app that combines: Media server, crawler, downloader, subtitle, cloudflare bypass etc, with plugin support.
This basically all comes for free in woltspace [0]. Woltspace believes that you should no longer bother building “scalable” apps, but personal apps that literally only run on your own machine. You can access them from anywhere (cloudflare, tailscale eg) and continuously improve them with your wolts over time - from anywhere.
It’s all open source and harness agnostic [1]. I’m the park ranger for life aka maintainer of the project. Did a recent YouTube video with what i think the future of open source multiagent setups will look like [2]
I've been running my instances for over 5 years, I update the Dockers every few months and that's it. I have about five users in different households that rely on it it has never missed a beat.
I’ve had mine going for what feels like ten years now no issues. Even longer if you could some of the earlier managers out there. The arrs are some of the most solid apps I’ve run. Especially once I switched them to use Postgres.
I think it just sounds intimidating if one hasn't tried, but they are really well designed pieces of software, that are relatively simple to set up (just have to go through the motions and connect everything with API keys).
Highly recommend them to OP. My stack went down once in the past 4 years and it was my fault.
I build my own version, single binary that does every thing. While I built it, I learned I cannot make it open source or distribute, bc it might be illegal to do so, so I assume many people have built that, but you're always locked into the narrowly scoped lego pieces. So I enjoy my media now, without hassles, all the features I need, none I don't. I hope one day I get to share.
Same. I made the mistake of adding the downloader to it at the start but brute forced my way through. Single Rust binary for all of the *arrs + sabnzbd is nice though. Unfortunately it is not something I am comfortable making public due to the all-in-oneness of it.
Almost all of the time went to to the downloader (usenet is more complicated than I thought). If I had just focused on the basics (sonarr, radarr, seerr) I could probably could have knocked it out in a day or so.
It is important to point out that “a day or so” - or even the ability to spin up your own version - is only possible because these projects already exist.
Without a tested and validated reference implementation that already accounts for the numerous edge conditions and quirks of media releases, APIs, etc., your vibecoded version would have taken much longer to iterate on.
ianal, but presumably you could share a research document that explains the theoretical kinds of broad strokes of an all in one type system, and how youd verify it works.
if anyone decided on their own to put that it into a local model and generate their own version of something that met the theoretical spec thats their own deal...
Oh cool! I have been running it happily for a bit over a year but found the interplay to be annoying. Currently working on https://github.com/tclancy/lintarr to try to "prove" rules across the stack don't conflict with each other.
The bigger reason is Sonarr and Radarr opens you up to a plethora of attack vectors and realistically at this point you're probably already compromised.
I don't see how. I run them with no public ingress, no plugins or mods. I don't torrent. I'm not sure there's really any attack vector worth worrying about. Certainly no more than many of the other services I host.
what the hell are you smoking? even the silly article earlier had the caveat "Yeah it downloads EXEs but its up to you or your family to run them"
and that's totally ignoring the fact that it's only a supremely naive user that hasn't yet refined their filters to exclude dangerous extensions (or, if they're too lazy to do that, rely on one of the many many pre-assembled filter/grabber settings across the net which already do so.)
what fear-mongering and FUD-spread.
You may also want to warn the public about the dangers of having curl, wget, python or gcc installed.
They expose a PLETHORA of attack vectors! you're probably already compromised, realistically!"
... and they're actually more dangerous than a free-floating EXE.
Shameless plug for my web video player that is especially designed for self-hosting situations. Drop a single html file in your web server folder with your videos and you get thumbnails, library, search, subtitle support, etc.
No required third-party dependencies (just the optional Google Drive and OneDrive features). Highly customizable (colors, site logo, default features) with a JSON file the player can produce. Responsively scales for any web browser: laptop, mobile, smart TVs, game consoles, even built-in in-car tablet browsers.
> I prefer running services natively when possible, but will containerize complex services or those with many dependencies. Love: a service that’s a single Go binary. A service that only comes as a Docker compose file and wants to run MySQL, Redis, and three other containers? I’m already looking for an alternative.
Ok, I respect your opinion, but remember that there are poor souls like me out there that actually love the opposite.
A single binary accomplishes most of those things just fine. Only thing it doesn't get you is isolation, but you can isolate things if you want even then (for example, FreeBSD jails are dead easy to set up).
ok, so you're just using a container with a different name? You can put the single binary in a container and call it a jail if it makes you feel better.
Well yes, but that's the point entirely - you can easily put a binary into a docker container, or a jail, or on a VM. But when something is shipped as 5 different docker containers it is MUCH harder to extract it into something you can run without docker or in a different configuration (like sharing a postgres instance between multiple applications).
One distribution method is flexible and can be deployed however you want, the other prescribes solutions and makes any customization difficult.
I'm confused. The original post said "why docker compose? a single binary dominates any docker compose", (I assumed dominates here means wins over because it wouldn't make sense otherwise?) someone replied giving the additional benefit of containerization, then someone replied to that saying you get all those benefit from single binaries, and if you want the system isolation you can use a freebsd jail, which is like saying "if you want the benefit of containers you can use containers"
Most applications do ship in a single container. whether it's a single binary inside, or a bunch of python or js files or shell scripts is a separate issue. If there is a dockerfile then you know how to build the container. a single binary is easier to build the container for yourself, assuming you also know the binary dependencies of that single binary. Like maybe ffmpeg or yt-dl cli on the path?
Applications that ship as a list of containers are a completely different problem. Those applications are either a victim of their own success, or have developers that come from a particular background.
I've never felt hindered by docker. It's pretty flexible. If I need even more flexibility I'm reaching for k8s, where I am still containerizing any single go binaries
Are you saying it's hard to edit a docker compose file? Being one service or 5 doesn't really make things more complex. I often point servers at my own postgres container instead of spinning up an extra one (so just remove that from the compose for example). That's kind of the entire point of compose, it's so flexible.
All my home server apps are in /apps/app-name/ all have a docker config... All data and configurations are mounts under that directory or to my library. It's really straight forward.
I don't have to hunt for different configuration files, each app has what it needs.
I can spin down, move and get back up in a minute or two. All in a consistent way to the platform.
The overhead for Docker isn't near a full vm, it's maybe 2-3% if that.
I don't use Docker but I love reading the compose file when a project has it. It is the quickest way for me to get a high level idea of this project's architecture. It's the best documentation :)
This. Docker compose is so great that you don’t even need to use it, for it to be useful. It can describe so well the dependencies your system needs (alongside proper documentation).
It’s the unified interface for me. Updates, logs, service management, network routing, config and storage mapping are all uniform and I don’t have to wrangle many slightly different implementations of each. Isolation from installed packages is a nice bonus.
I like to have consistent architecture across all my services, and I like to see and configure everything in (mostly) one place. If you're running a single go binary, sure do whatever, but if you're running a go binary, and a *arr stack, and a bookmarking service, home assistant, frigate, etc, it gets out of hand pretty quickly.
You trade a bit of upfront setup (I mean barely now in the age of LLMs) for easier automation and flexibility down the road. I use docker compose to run my local dev stack and qa envs, but then can do whatever I want with the containers for prod. Google Cloud Run, Amazon, whatever. Right now I do just actually use docker compose behind nginx in prod, but I've used basically the same docker compose setup for all my projects that aren't large distributed systems in the last 10 years. And when the system gets too big, I refactor it into k8s, but k8s is overkill for like 99% of my personal projects and like 75% of my professional projects.
for long-running services, I've never taken the time to learn something reliable like systemd. with Compose, I can see everything I need to know in one yml file and the commands to go with it. I also don't have to really understand any dependencies or their own configuration and can cleanly remove everything associated with that service with a few commands.
You can use systemd and docker compose together. A systemd service is a single config file that runs the commands you give it, including `docker-compose` commands. Everything else about your service is still orchestrated via docker, and you can still run commands against the running containers just like before.
A big win with very little effort is systemd starting your docker compose service on boot. You have your systemd config file set `ExecStart` to call your docker-compose command, throw the config file into your systemd folder, and tell systemd to enable it. Docker compose handles the rest.
Not OP, but i prefer a single binary on docker, at least i know it won't fuck up my filesystem when i do something stupid with it.
Or at least i can reinstall any time i want, because i know it won't leave any trash in my filesystem and when reinstalling, it won't reuse those outdated files
I’ve been a die hard native package or static binary proponent for a long time. I still hold that there are real advantages.
As podman has improved and compose stacks have improved, I’ve conceded that containers simplify deployment and updates and isolation, and at this point I find for me the benefits more than outweigh the hassles.
Instead of crafting a set of ansible plays that do all the manual native installation steps, I have a relatively lightweight ansible rig that reads containers and relationships desired from a manifest in inventory, then builds or modifies compose files, then just deploys. Much less futzing. And I’m more likely to keep things up to date regularly as a result
Or, increasingly, dispatch codex to handle it for me
In this way I update home assistant across multiple instances now like magic. I don’t even dread it a little anymore
Some isolation is nice. I say some because I’m told it’s security is not to be trusted, but I still like that even a non nefarious binary has ootb boundaries on what state will persist.
Docker isolation is likely good enough if you keep your kernel up to date and aren't allowing 3rd parties to run code. If you are running a CI service or something you need micro VMs. But to host your nextcloud server, docker is good enough, it's unlikely someone will both exploit nextcloud and the kernel if you are keeping them up to date.
docker isolation is nice just as a management boundary. It's nice to be sure applications don't accidently shit all over each other. Like an application that decides to clear /tmp because reasons, or 2 random ones that both store temp state in a /tmp/state.dat for some odd reason.
My favorite is applications that conflict on ports. So you go to change the port an application is listening on, the docs say PORT env var so you set it, but it doesn't work.
You look through the code (and this is before LLMs) but the config is just a labyrinth of values flowing from god knows where. So you search the repo for PORT and eventually find a PR from 2 weeks ago that changes PORT to ENDPOINT, so you set ENDPOINT, but it still doesn't work.
Then you find an issue that was opened last night that the new `ENDPOINT` value doesn't load from env vars because it conflicts with another ENDPOINT_ prefix on another value, instead you need to create a config.toml and set an `endpoint =` in that, so you do that.
Then you realize that creating an empty config.toml has now overridden all the previously defaulted values into nulls or empty strings, instead you need to run `foo config-defaults > config.toml` and go edit the endpoint one.
Only to realize that there is a secret that the tool uses to encrypt an ephemeral session. if you don't specify one, the tool would auto generate one on startup. The only downside is restarting it logs your client out which isn't a big deal. But now that you switched to config.toml you must set some value there and it can't be empty
Then you're just wondering why the hell did I want to change the port?
I too prefer the Docker compose route. If you give me a single binary, I'll go out of my way to run it inside a compose stack.
I manage my home server with a script that presumes the service is running on a compose stack. So I can run stuff like `stack freshrss restart`. I don't want to touch my scripts, I just add the compose file to some expected path and the script finds it based on naming conventions.
I would rather keep things on postgres but everything wanting its own postgres container is really annoying. Postgres already has support for multiple databases, users, and access controls. I would personally prefer to just set up one postgres instance with proper backups and have everything run against that, unfortunately some self-hosting projects have fairly poor support for that.
You could sometimes get there. The problem is in an effort to simplify the getting-started with postgres (or other databases) most projects assume they would also automate the management of that postgres instance.
The getting started would be a lot different if it first instructed you to install postgres, run a script or create a database and users and configure access controls. Also certain things become easier if you assume you can also manage the postgres instance, like on an upgrade, maybe you just create a completely new database and drop the old one, or isolate tenants within one application into different databases, etc.
In general, I agree with OP that sqlite is all you need for selfhosted applications (most of the time at least)
That's a large part of why I lean more and more into just preferring sqlite for applications.
Even in work environments for apps that likely won't have thousands of simultaneous users. Modern SSD drives are so fast that so many apps don't need the advanced features of a full blown rdbms. And I think options like Turso and Cloudflare D1 fill the in between as a hosted option.
I do wish that sqlite had a couple more native types, it works really well as it stands.
> several things each wanted their own Postgres container. Immich wants a Postgres and a redis instance.
Those things are dirt cheap. A Postgres container on my machine uses 200MB of memory. Redis is nearly free. It's a small price to pay to not have services cross paths with each other and mucking everything up.
My experience with this is that all of the projects that are worthwhile, continue to grow and then at some point some kind of bottleneck is hit. Migrations become slow or overly complicated, lookups start taking forever, etc and the app just starts to slowly become unusable the more you try to use it.
You end up in a situation where you are following some complex guide (hopefully written by the actual developer) where you make a huge backup, have to tear everything down, and then hope the migration from SQLite or MariaDB doesn't hit any issues. Postgres is heavier but at the scales of self hosting it should not matter on any modern system. Multiple docker containers even on a Pi isn't something you have to worry about. Solutions like Immich are usually run on a NAS where there would never be an issue
Also you don't always know who your users are gonna be. You could create a cool personal photo backup app for your <20k library, then get people with a dozen terabytes of photos wondering why is it so slow?
Good article. I have many of the same services in my setup.
Self-hosting email fully is tedious/unnecessary, but what you can do is mirror your various inboxes (Gmail, iCloud, proton, non-corporate outlook) with mbsync and use roundcube or some other frontend to view emails and compose.
I use this to easily access mail from different services, and I can also send from any of them.
I also have a seperate service that watches the mail directory and sends a ntfy.sh notification for new emails.
Certainly not unnecessary if you care about privacy and Google not reading your mail.
Is it tedious? Not really for someone with a basic command of Linux and computer networking. Even in the pre-AI days, it only took me an afternoon to set up Postfix/Dovecot, and a few more days to refine various aspects of the setup and to write a bash script that can rebuild the whole thing on a new machine. There is basically no ongoing maintenance at all. The Dovecot syntax changed in Debian 13, and Claude translated the old config file in less than a minute. If you are willing to use a Dockerized solution, it's even less work, but I personally like to run this kind of thing bare metal and control every aspect of the setup.
I self-host e-mail, because I am stubborn - I have done it for 20 years. I would not recommend anyone do it if they aren't already, or if they well-and-truly want to be on call for their homelab.
In the last 20 years, I've also run into problems like:
* My brand-new ISP wouldn't allow TCP port 25 inbound, and wouldn't relent. I had to go back to the old ISP.
* SPF became mandatory at Gmail, but I didn't get the memo, and Google started bouncing my mail one random day.
* My residential IP ended up on a "policy blacklist", one it could never get off because it's blacklisted by virtue of being a residential IP, not that it did something bad.
I've never tried to set it up even though I know it can easily be done with things like mailcow.
The downside is never really knowing if your email reached your recipient (maybe it got blocked by Gmail because of your IP, maybe some other reason). If I use Gmail/proton etc, then I can be pretty sure that the other person got my email.
You'll know from bounced email and logs, which you should have monitoring setup on.
It's not trivial to run your own email stack, but it's not hard either. It's more tedious than anything. Once it's up and running, it's low-maintenance. And it's fairly trivial to avoid spam. I get vastly more spam in my Gmail than I do on my self-hosted domains.
The biggest challenge with self-hosted email is getting an IP address that's not blocked or carrying low reputation. Most VPS IPs from e.g. Digital Ocean, Linode, Hetzner, AWS, etc. are going to get blocked from Google (the strictest) and Yahoo, Hotmail/Live/MS, etc. Cloud providers often block incoming and outgoing mail ports, too. I have a Digital Ocean VPS and my account is old enough to be grandfathered into the old policy, which means my email ports aren't blocked. I used to have no trouble sending email anywhere from my VPS, until I downsized it and got stuck with an IP in a bad reputation CIDR. Now I use smtp2go for relaying outbound mail, but still handle all inbound mail. 10/10 score on mail-tester.com, and I never have any problems or need to do any maintenance.
A few major vendors (I want to say Microsoft was among them) were infamous for silently dropping email without bounce backs. They'd perform then whole protocol sequence as if they accepted the email, then simply never actually put it in that user's inbox or even spam folder.
People should read the parent post and stop spreading FUD about email. The only part that isn't easy is the IP requirement. If you don't have a fixed public IP, preferably v4, then self-hosting email is out of reach, but so is hosting other things like game servers. It's the main cost of self-hosting, and one that isn't big relative to all the benefits. Privacy is the biggest one I think, and so is the ability to create unlimited aliases, one for each company/website.
Because it's not about setting it up, it's about not getting outgoing mail perceived as spam by someone. Borderline impossible without scale. Even doing everything correct is not enough.
> . I do have a 4k TV, but it’s not extremely large, and while I can tell the difference in resolution between SD and HD, for almost everything, 720p seems to be good enough for me.
I am actually jealous of you, I can tell the difference between 4k low bitrate and 4k high bitrate, 1080p and 720p are like mud to my eyes.
I got a VHS player and a load of free tapes. It hasn't changed since I was young (well they've degraded a little), so if I enjoyed it then I can enjoy it as well now. In some things I try to resist a little the encroachment of higher standards.
But not in everything. I can't drink a cheap scotch.
It's been really interesting seeing the divide in the comments! I have a completely different philosophy from the OP on how to self host haha, and it comes from not wanting to fiddle with my server all the time! But they say our ci/cd is fiddly! The whole point of it is to NOT be fiddly! People gotta start figuring out how things work instead of throwing spaghetti at the wall.
In my experience, when something involves words like "declarative", "deployment", "docker", "k8s", that's a sign that it will actually involve more fiddling rather than less. But that's just me.
I'm running like 10 services on a VPS. I'm a professional software engineer. Do you think I want to ssh into my server and find a bunch of single binaries all over the place to run them? What if a service stops, how do you automatically restart it? How would you automate creation of a qa environment when you create a PR on github so that you can test your changes before they go to prod? Gonna ssh into the server and pull down your repo and build the app and run it?
It takes like a day to figure out docker compose and like a week to figure out k8s at a level where you can get things done. Honestly, you can get something working now with LLMs and have them help you understand it in probably less time.
Of course I want a declarative deployment for my complex home server. Do you run like 1 app on your home server? It's not "fiddling" if you know how these things actually work.
I imagine it's more fiddly to figure out how to wire up all your individual services to Uptime Kuma and Homepage. You don't even need Uptime Kuma with docker compose, instead you just set a container restart policy and have a healthcheck endpoint.
Pets are OK sometimes! They do need care and feeding, though; some of us don't see that as chores, instead delightful hobbies. I can definitely see the perspective of "there are better things to do", especially if you don't like those things.
There is something satisfying in handling a home server as cattle though!
I don't do it because I don't care; I do it because it is super satisfying when it has been boringly running for weeks without me doing anything to keep it up beyond paying the electricity bill, upgrades are extremely predictable, etc.
Especially now when you can snap your fingers and get a semi responsible codified version. Heck just do what you want to them ask the LLM to port your work to something marginally better than a totally hand cobbled bucket of bits system.
That’s what I’ve been doing. Managing this stuff by hand is awful and I’m not a sysadmin who is good at it. I just added single sign on to my home network for a wide variety of various services. Some of them didn’t even support SSO but a quick AI patch fixed that. There is no way I would even attempt that by hand let alone succeed.
I saw statements to the effect of “this would be nice but it’s a lot of work” several times in the post. I don’t understand why that has to be a blocker anymore. All this tedium is gone.
I see people say the same things in threads discussing open source agent harness repos that are MIT licensed. I don't understand how it isn't obvious to them at that point haha. I don't think people realize they can just... do whatever they want to
Pfsense and maybe openwrt allow split horizon DNS that allows Wireguard tunnels to work seamlessly whether you’re on or off your home network. Took me years to figure out but it’s what I use now
foobar's rock solid. I have a portable copy on a mounted NAS share and it's great for transcoding and being an ad hoc music player while setting up new computers
Just a note on OP's hardware choice. I used to have a Beelink with the N100 CPU. That thing overheated and reboot intermittently after couple months of usage. I suspected they overclocked the CPU to look good in benchmarks. I tried to underclock in the BIOS. The temperatures dropped but the setting was reverted after the next power down. Nothing in the BIOS stuck. Their support was worse than useless. The machine ended up as spare parts.
You are better off getting the used business mini-PC from HP/Dell/Leveno/Asus. They have better specs and run great.
This. Or just buy an N100 motherboard (Asrock makes one) and put it in a standard ATX case with a proper cooler, power supply, and enough space for expansion cards and hard drives etc. Maybe you will want a GPU in the future, even if it is unncessary now. Small nonstandard form factors always come back to bite you sooner or later. It's better not to cut corners and build something modular.
Most likely something wrong with the cooler, maybe it's not making proper contact with the CPU. Should be repairable if you give it a try, but yes, the used thin clients are higher quality.
I did clean the thermal paste between the CPU and the heat sink, and re-applied new thermal gel. That didn't help. It was the raised voltage preset in the BIOS. If I lowered the voltage, the temperature stayed low and it would run fine, until the next power down reverting the BIOS changes. The box was too small, the heat sink was too cheap, and the fan was too lacking to be a proper overclocking setup. It would run fine out of the box for couple months until the CPU and hardware worn out due to stress and heat.
Hmm that's too bad. I have bought a few of those (different models) over the years to use as HTPCs for different family TVs. Only one had cooling issues (bad fan bearings). I avoid the extra small ones and just get the NUC-like ones.
Just another data point, I got a Beelink S13 Pro N150 running similar things as OP and haven't had any issues. I made changes in BIOS (set "Chipset > PCH-IO > State After G3" to S0 so it boots up again after regaining power) and they are sticking for me.
Same here, N150, no issues. Don't think I've made any changes to the BIOS.
Fans spin a lot near-constantly though, I think it is the on board M.2 SSD that is hot, I need to get around to adding a thermal pad or something one of these days.
I run Adguard on my router, not on my server. I did write a subsequent post about the rest of the stuff I run on my router and my Proxmox cluster, but forgot to mention Adguard.
With some small systemd configs each native app is essentially a container these days. The overhead of unique kernels per app would really restrict what could fit on a small box like this
Like the other person said, proxmox allows containers or full VMs, and the base OS takes less than 2 GB RAM.
My jellyfin LXC idles at a few hundred megs of ram, shares the base debian kernel, and if anyone does ever break in, they don't get access to anything else, anywhere else on my server. (All media is also mounted read-only to the jellyfin LXC.)
I really, really like it, just set it up recently. Everything in its right place.
Heya, nice read. I’m wondering how you managed to get LetsEncrypt certs minted when your caddy isn’t exposed to the public internet. Doesn’t LetsEncrypt need to make a request back as part of the handshake to mint/refresh the certificate?
Would love to know as I wanted to do the same in the past and ended up self signing certs.
As someone that loves docker compose and doesn't mind services with the benefits of software better than sqlite, I guess I'm the minority here. I wish more apps were based on postgres. I also have never had a problem with my publicly exposed applications. Not once in 8 years.
Nah I thought their setup was pretty ridiculous and seemed hard to maintain. I run all my stuff via docker compose and am very happy with it. I don't understand the comments praising single go binaries, which I've actually seen reach out to the docker engine api and start containers for their dependencies
Owntone holds your collection on the sever and can stream to airport, airport2, dlna.
Shairport-sync turns a raspberry pi zero 2w, or orange pi zero 2w into an airport {1,2} speaker. Or use a (super cheap) esp32 airport 2 from aliexpress.
Can sync every room in the house with owntone OR "I want my phone or laptop's sound to come out of the speakers in this room"
Owntone has access to your paid spotify account if you want that. Can stream to your phone with http (remotely with wireguard vpn). Also has mpd controls if you want those and can use apple remote app on your phone if you like that.
Owntone, pihole and wireguard is reason enough alone to put an old laptop into service as a home server.
I have gone down the self-hosted rabbit hole, like many it all started with setting up a pihole on a raspberry pi.
I have since gone down the jellyfin + arr stack route. Resource wise the entire stack uses very little. I have Sonarr, Radarr, Prowlarr, Bazarr, FlareSolverr, Jellyseerr, SABnzbd, qBittorrent, Gluetun, qui, Decluttarr, Recyclarr, arr-dashboard all running on my dell optiplex (previously on the raspberry pi).
All up this uses about 2GB of RAM and 0.5% CPU.
I found it extremely fiddly to set up, understand what each component does and have everything integrated well across all my devices.
Without the help of an AI assistant I just wouldn't have had the time to debug every little issue. It took me a good 12 months of tinkering around work hours to have everything stable. However now it runs along fine with very little intervention and I'm able to share it with friends and family.
So the resource argument I don't think is valid. However the fiddly mess I do agree with. Although personally I found it fun and rewarding when things finally worked. But I can understand why people would not want do spend all that time on it.
I'm puzzled that OP seems stumped on syncing keepass when they're already using syncthing. I've been using keepass+syncthing for years, and I've never had any issues.
I considered KeePass+ Synching but tended to run into the same issues as with Owntracks - Android is hostile to always on background apps. Then when it came time to onboard my spouse, KeePass wasn't an option.
I recently bought a Lenovo Mini PC (01Q8X10) for under £300 that has a Snapdragon X1 CPU (and 32GB LPDDR5 RAM!) and it's turning out to be perfect for self-hosting running the newly released WSL Containers on Windows 11 with WSL 3 (it doesn't fully support Linux yet). Podman had some issues with WSL 3 that they're still releasing fixes for, but it's all working with that too.
I don't know why so many people host personal forgejo / gitea instances. You only really ever need a server with ssh access (and cgit if you really want a web interface).
Just FYI you can get a M1 Macbook Air with a broken or cracked screen for about $250 on eBay. It's a headless device at that point but it has a keyboard and touchpad.
I use it for my Home Assistant setup and more. It still has headroom. No fan, very low power consumption. Not true Linux container stuff but pretty much everything works on it in this area.
My go-to for low power consumption is a Chromebook running some vanilla flavor of Linux. I have some HP units that will run for almost a day on battery with the lid closed. $25<$250.
How do they do with staying on and then restarting everything after power off? Can they automatically repower?
I don't have my embedded macbook fully configured but gemini says I can get it to repower itself.
By repower I mean if the power goes out and then returns the machine needs to power on and boot.
I have a couple of older Chromebooks I use in rando locations (basement, shop, car, rv, boat etc). And they are great for that casual scene-specific use. Very much recommend and amazing what you can get for sub $50 easily.
One can create a gmail account like my-robots-73489@gmail.com to sync everything too without touching your main google account. Very handy.
[edit: one thing about the M1 is that it has native h.264 h.265 and prores hardware encoder decoders if you are doing video intake of multiple streams]
Great write up, thanks for sharing. I run some of the services mentioned, but got some good ideas from the article and the comments.
One suggestion in place of MPD is lyron (https://lyrion.org/). I’ve used MPD for years but recently switched to lyrion and for me, it’s a lot easier and suitable when you have many clients. It’s nice to have a single (nice) UX/UI for my music and all the “speakers” on the local network.
I've been building and self-hosting my own analytics platform[0] for 13+ years and never had any major issues with it.
The advantage is that if I do want a specific feature, I can just add it, and I am never worried about subscription limits or surprise charged, because I host it on a fixed-cost VPS.
Things don't really need maintenance and don't really break by themselves, the only time when they can sometimes break is when changing things.
You do need to have a proper setup though, for stuff like disk space alerts, knowing the common gotchas of disabling settings that can cause issues, etc.
They might have to turn on/off some lights using the buttons on the smart switch, yes. Or maybe lose access to Jellyfin for a while. But on whole things have been pretty reliable.
How do people use let’s encrypt on a host that is not accessible from public Internet? Do they manually renew every three months through DNS challenge?
Podman + systemd is how I run production workloads at work.
For us it's the perfect balance between simplicity, observability (Beszel) and reliability. Not the right tool if you have dynamic workloads but definitely the right tool for boring and predictable workloads.
At home I run K3S, it's awesome. It also breaks a lot and does add a whole bunch of complexity that I would only recommend if you enjoy tinkering and fixing things.
I second this, it's great. Commonly called "Quadlets".
I also use BGP advertised pod networks so each pod gets it's own IP address that's directly routable (l3 ipvlan) with the rest of my network (hint, don't use 100.64.0.0/10 for that if you use tailscale), so I don't bother with NAT on the host itself or port mapping or port conflicts. It also makes caddy in front of it super easy, barely an inconvenience.
I prefer docker containers (on proxmox VMs), but that could just be the familiarity of a few years experience talking. It does add some distance from 'bare metal' but it also makes backup/restore, and therefore portability, quite simple. I backup both the VMs and the 'persistent' container data that I always create dedicated volumes for, so I have a couple of different avenues and granularity for restoring.
There's nothing else out there, that I know of, that offers improvements on any kind of scale that it would be worth my migrating to.
I do hear good things about podman, so if I was starting from scratch I'd look into that. I have one podman instance running UnifiOS, but that's only because there's no official docker compose from Ubiquiti (and maybe that's telling in itself). Podman feels more opaque to me than docker, but I have years under docker and essentially zero under podman.
I had an easier time moving a project from docker compose to k3s after wasting so much time with podman compose (maybe expected considering than k3s is still root daemon)
I personally use NixOS and a mix of systemd services and Quadlet-Nix for the few containers that have too many dependencies and moving parts. Works well.
Container manager on Synology. Don't have much choice. Some stuff can run native but it's easier to run containers.
The big issue with Synology is the really crappy db logs driver they use. If a container has been running for a while it takes an age to see the logs via the Web UI.
There are really easy to user docker app managers out there. The author says 'I already use systemd'. I mean, I guess I'm using systemd, but really I'm using e.g. runtipi with an app store, web-based config and logging, and a super easy app template. I can point claude at any project with a docker release and in a couple minutes have it show up in my app store to install, run, simple data backups, etc.
Docker containers are fine. I've moved between 5 NAS' over the years and docker has enabled it to be seemless.
rsync data directories. docker compose up. Done.
On the consumer or even prosumer nas products which have a semi-locked down system (ie Truenas/qnap/Synology). Technically you have root but as soon as you update the OS all/most of the changes end up being broken. Yes you could probably do something and deal with it. But personally I've spent too long babysitting it on the first and second nas that I no longer care.
Quadlets rule, I have tons of them configured for my professional and hobby/self-hosting services. Super convenient, I wish there was something equivalent for my Mac (launchd files just don't cut it).
Most of the time projects provide a ready to use compose file. I use dockge that provides a simple web interface for compose configs, so just paste in the compose file there and run it.
I am surprised author said *arr stack is a mess, it's probably one of the few things I haven't touched much since setting it up and it saves me a ton of time managing new media.
People twist themselves into a knot talking about AI as the future. Its more of a bubble waiting to pop. Self hosting is the real disruptive future.
The hardware to accomplish most things that we rely on from cloud vendors is absolutely trivial. My own home server is a used super cheap and tiny machine. The hard disks don't even fit in the box and yet it has plenty of CPU and memory to spare.
The software is getting easier to configure year after year eventually creeping to the point where elderly people too lost to use a remote will be running this stuff. The real challenge will be solving for easy access to private networks from public spaces across the internet. Right now everybody seems to be centering on vendor solutions like TailScale for this.
I self host (I have a large overlap with the services in TFA), and love it. I will never go back and think that more people should self host (or at least...join small groups of people who designate one person to manage it for a small number of people).
And believing that the current AI economy might have a period of rapid devaluation in the near future is pretty reasonable. But, much like the original .com bubble, the idea that AI as a technology will not survive and thrive through such a pop, and become an ever more important part of our future, is a belief that I can't comprehend. Anthropic and OpenAI very well may go belly up. AI as a technology isn't going anywhere.
And as much as I'd like to believe that more people may self host more things in the future, self hosting will never, even if it was adopted by literally everyone, become as big a deal as AI, even if AI hit a wall tomorrow and didn't get any better than the current frontier models.
Self hosting has never been easier and I imagine the average person has never been further from trying it. No one will ever get anywhere close to hosting this much sruff
Self hosting is mostly a hobby like home brewing beer or woodworking. It’s more accessible than ever but there’s still little incentive for the average person to do it.
The average user is much safer putting their photos on iCloud than a home server unless they are tech inclined and personally invested in keeping it secure and backed up.
I got into it about a year ago. Bought a refurbished old Dell mini PC for 300 bucks and turned it into a home server. Debian + several docker containers.
I don't know if this is the future, but it feels disruptive. I think it is amazing that I can run my private home streaming service alongside a bunch of other cool things on old hardware and the experience is in many ways superior to actual services that I am charged to use.
I wanted to self-host Jellyfin, but hit the problem.
It's bad at the main thing that player should offer - media playback.
If you open https://jellyfin.org/docs/general/clients/codec-support/ you can see that this doesn't support HDR and that doesn't support H265.
There is Jellyfin Desktop and it breaks when display scaling is over 100% (what year is it?). Some clients have atrocious subtitle support. And then none of them apparently can't properly pace frames so I still see judder as if I'm on 60 Hz screen (24->60 can't be even translation). But my screen can do up to 165 Hz, so that should never be the issue.
In then I just went back to playing media files from Samba share with MPC + madVR. No automation or catalogization or anything, but it is supreme at actually displaying things, and that's what it's important to me.
Fair, I just never use the web clients. I mainly use the Android phone or TV apps, which do support H.265. The Android TV subtitle support seemed ok, I think I did have issues with the Roku app and subtitles.
I understand use cases for Jellyfin and selfhost it myself for music (in combination with incredible Finamp[1]), but for movies/shows plain MiniDLNA[2] works like a charm and is accessible from any device.
I used jellyfin-mpv-shim. You select what to play from any ui and it starts playing in mpv, which seems to have excellent support for everything you would want a player to do.
Tried it too, it still had judder as all other clients I've seen. I think it's either server software problem or maybe all clients fail to do this properly.
After Google Assistant died I immediately installed Home Assistant on my old PC. Not as good, but at least it doesn't take 5 seconds to turn on a light like Gemini
I stared self hosting in 2020 when Raspberry Pis were super cheap (not anymore :( ) and lots of things started becoming available in Docker. Haven't ever considered going back to the cloud. I love taking back full control.
Same, Covid was a great time to spend shifting things from clouds to home. I have a small fleet of Pi4's and 1 newer 5 (for running Omada locally). Cost has shifted from subscriptions ($$$) to maintenance (time) :)
Out of curiosity, do you have any concerns about backups (especially for e.g. VaultWarden and Immich)? I've been pondering running them, but figure that I'd need to back up to The Cloud for me to feel less concerned about losing everything, at which point: what's the point? Curious what you're doing on that front!
I use both Restic and Syncthing. I have considered putting encrypted backups on a cloud blob storage, but my preference would be to set up an offsite backup at a relative's house.
I use restic to create encrypted snapshots on external cloud storage for the data I care about. My media is stored on a nas which covers me for drive failures. If I accidentally delete everything or my house catches fire I can download my personal data again and just redownload the terabytes of media.
About to retire the NUC I used for 10 years running arr stack, jellyfin, home assistant, pi hole and a few more containers. Bought a new bee link mini pc, new Intel CPU, 500gb SSD and 16gb ram for less than $500.
The best thing to self host (if you have some international family/friends you trust) is a VPN. Get each household a raspberry pi joined into one vpn. It means you actually get local tv to work (unlike Nord vpn etc). Each pi sits on a guest network so traffic passing through can't touch anything in the house
I have been self hosting for over a decade and recently have seen more people in my circle doing that and I have been helping them.
At first it was hard but then i made it into a product. I am working on Tarvis.io which makes self hosting accessible to non technical users as well. It comes with installation assistant, ssl domains and backups inbuilt into it.
You can install open source apps from a catalog or drop any github repo and it installs and run it for you.
With Tarvis, I have been self hosting more since I am able to install and delete apps with one click, it makes it easy to try things.
> Sonarr, Lidarr, Radarr, etc. Every time I read about setting it up, it just sounds like a resource-intensive mess that’s fundamentally prone to frequent breakage. That’s the opposite of what I’m looking for when choosing self-hosted services.
I just checked on my *arr stack. Sonarr and Radarr idle at 157MB and 159MB of memory and approximately no CPU. Jellyfin currently idles at 330MB. I'm not sure where you got this impression from. I grant these tools are complicated and have somewhat clumsy UX, but they aren't especially heavy to run and are not fragile.
OP, point your :robot: at https://github.com/subdavis/selfhosted/blob/main/docker-comp... and ask for an explanation. It's a single compose file that illustrates how the *arrs, transmission, and jellyfin all work together. I'm sure you can set it up :)
Indeed. The other thing to point out is that, once you have set them up, they will chug along fine for years with near zero input.
Here is the stack I recommend:
* Radarr (movies) with optional second instance for 4K versions
* Sonarr (shows) with optional second instance for 4K versions
* Sonarr (anime), separate instance to have control over quality settings
* Prowlarr (tracker/indexer management)
* SABnzbd (Usenet)
* qBittorrent (torrents)
I would also recommend consulting the TRaSH guide to tune your setup and ensure you’re grabbing quality releases. From there, the next logical step is Profilarr :)
sabnzbd can be replaced with nzbdav or similar tools (infinidysk, altmount) if you want to free up storage and stream directly from usenet.
Or, even better, the entire arr stack can be replaced with stremio to just keep everything on-demand instead of curating a library at all.
“Even better” is relative. Some people like having media stored locally instead of relying on torrent streaming availability.
Usenet streaming sounds cool, but with releases being DMCAd so frequently, I don’t see how it would be practical for anything but the most recent content.
Also, I imagine that Stremio without Debrid (or equivalent) is kind of a terrible experience most of the time.
I tried it once and it barely worked at all. Plex would just show everything as unavailable. Really awful experience even when it did work.
> stremio
Is this the latest rendition of Popcorn Time?
Yep basically. The real magic is when you combine it with a Debrid service, which essentially puts a cache in front of torrents. If the torrent has been downloaded before, you get a direct hosted stream instead of having to stream your movie from active seeders (i.e. buffering hell).
That sounds slick. I'll have to check it out!
Where does that hosted stream live? I'll have to peep their docs. Sounds interesting
The interfaces of arr stack are ugly, and linking qbit to indexer manager to arr app is needlessly complex. Maybe there could be plugins in one app like qbit.
I used to think so too, but now I usually just have Muse or another AI agent on my box do those things for me. I should maybe create a slimmed down API first *arr stack equivalent then hmm
Curious question, I host a single instance of sonarr that supports anime through tags. Is there an advantage to hosting 2 instances instead?
Single instance works fine, but a separate instance allows you to set global quality settings that are more accurate for anime: https://trash-guides.info/Sonarr/Sonarr-Quality-Settings-Fil...
You can configure quality profile via the tags too (I do have separate profiles for anime vs liveaction). I didn't even set the tag, as I don't intermingle them in the same directory, and automatic tagging by sub path is possible too
So I'm also surprised why someone would configure a second instance for that
If using the TRaSH guides (synced automatically with recyclarr), with all their quality profiles etc, I find that those mostly don't make sense for anime. It's easier for me to keep the anime specific stuff separate, different naming conventions, different episode numbering, etc.
Can you explain? Are you manually setting tags for every anime you add? How does it work with requests made thru Seerr?
Prowlarr replaces Jackett?
https://github.com/Jackett/Jackett
Yeah, it interfaces way better with Radarr/Sonarr. I still run Jackett though, good for manual searches.
There's been two major disruptions in my experience.
First was Lidarr went for months last year (was it this year?) unable to search/add musicians due to complications with how they were relaying those searches. It took them a long time to refactor that part of the software and make it usable again.
Second was the retirement of Readarr, the ebook-version of this software, which was an absolute shame.
But aside from that this is pretty much set-and-forget software. Although the setting is harder than it should be, should be much easier to favour smallest x265 rips at <resolution>, lossless audio, season rips with subtitles.
In this day of vibeslop, it's insane no one thinks to combine all these into single app. The best we can do is having single docker compose file that has to manually point each service into each other.
A single dream app that combines: Media server, crawler, downloader, subtitle, cloudflare bypass etc, with plugin support.
This basically all comes for free in woltspace [0]. Woltspace believes that you should no longer bother building “scalable” apps, but personal apps that literally only run on your own machine. You can access them from anywhere (cloudflare, tailscale eg) and continuously improve them with your wolts over time - from anywhere.
It’s all open source and harness agnostic [1]. I’m the park ranger for life aka maintainer of the project. Did a recent YouTube video with what i think the future of open source multiagent setups will look like [2]
Happy to chat more about it
[0] https://woltspace.com/ [1] https://github.com/jerpint/woltspace [2] https://www.youtube.com/watch?v=H-4khO7o0XQ
I've been running my instances for over 5 years, I update the Dockers every few months and that's it. I have about five users in different households that rely on it it has never missed a beat.
I’ve had mine going for what feels like ten years now no issues. Even longer if you could some of the earlier managers out there. The arrs are some of the most solid apps I’ve run. Especially once I switched them to use Postgres.
I think it just sounds intimidating if one hasn't tried, but they are really well designed pieces of software, that are relatively simple to set up (just have to go through the motions and connect everything with API keys).
Highly recommend them to OP. My stack went down once in the past 4 years and it was my fault.
I build my own version, single binary that does every thing. While I built it, I learned I cannot make it open source or distribute, bc it might be illegal to do so, so I assume many people have built that, but you're always locked into the narrowly scoped lego pieces. So I enjoy my media now, without hassles, all the features I need, none I don't. I hope one day I get to share.
Same. I made the mistake of adding the downloader to it at the start but brute forced my way through. Single Rust binary for all of the *arrs + sabnzbd is nice though. Unfortunately it is not something I am comfortable making public due to the all-in-oneness of it.
Almost all of the time went to to the downloader (usenet is more complicated than I thought). If I had just focused on the basics (sonarr, radarr, seerr) I could probably could have knocked it out in a day or so.
It is important to point out that “a day or so” - or even the ability to spin up your own version - is only possible because these projects already exist.
Without a tested and validated reference implementation that already accounts for the numerous edge conditions and quirks of media releases, APIs, etc., your vibecoded version would have taken much longer to iterate on.
Solution is easy, in fact there are 2 solutions:
1. Clear that "might", it either is or isn't. An AI can get you 90% there, maybe the definitive confirmation would be good from a lawyer.
2. None of this matters if you share it from a foreign code forge and an anonymous pseudonym.
ianal, but presumably you could share a research document that explains the theoretical kinds of broad strokes of an all in one type system, and how youd verify it works.
if anyone decided on their own to put that it into a local model and generate their own version of something that met the theoretical spec thats their own deal...
Maybe the resources in question arent RAM.
Another good option is nixflix: https://github.com/kiriwalawren/nixflix (a nixos equivalent)
Oh cool! I have been running it happily for a bit over a year but found the interplay to be annoying. Currently working on https://github.com/tclancy/lintarr to try to "prove" rules across the stack don't conflict with each other.
The bigger reason is Sonarr and Radarr opens you up to a plethora of attack vectors and realistically at this point you're probably already compromised.
I don't see how. I run them with no public ingress, no plugins or mods. I don't torrent. I'm not sure there's really any attack vector worth worrying about. Certainly no more than many of the other services I host.
https://news.ycombinator.com/item?id=50108872
That’s been a thing for years. No one is actually running the exes, or at least they shouldn’t be. Its such a simple thing to prevent too…
Simply put, skill issue. Never ran into anything like this when using reputable private trackers and indexers.
Unless there's a bug where jellyfin or plex decides to execute a random exe it tries to index, exploitation seems unlikely.
And even then you should be running in a container at least.
what the hell are you smoking? even the silly article earlier had the caveat "Yeah it downloads EXEs but its up to you or your family to run them"
and that's totally ignoring the fact that it's only a supremely naive user that hasn't yet refined their filters to exclude dangerous extensions (or, if they're too lazy to do that, rely on one of the many many pre-assembled filter/grabber settings across the net which already do so.)
what fear-mongering and FUD-spread.
You may also want to warn the public about the dangers of having curl, wget, python or gcc installed.
They expose a PLETHORA of attack vectors! you're probably already compromised, realistically!"
... and they're actually more dangerous than a free-floating EXE.
> Sonarr, Lidarr, Radarr
There’s no value to this post honestly.
There must be large amounts of similar code between Radarr, Lidarr, Sonarr etc.
Whenever I use them I wonder if they could be one unified app with plugins...?
Also the barrier to entry for these things has dropped immensely with ai
Shameless plug for my web video player that is especially designed for self-hosting situations. Drop a single html file in your web server folder with your videos and you get thumbnails, library, search, subtitle support, etc.
No required third-party dependencies (just the optional Google Drive and OneDrive features). Highly customizable (colors, site logo, default features) with a JSON file the player can produce. Responsively scales for any web browser: laptop, mobile, smart TVs, game consoles, even built-in in-car tablet browsers.
https://github.com/pseudosavant/player.html/
Hey that's pretty neat. Thanks for making and sharing.
> I prefer running services natively when possible, but will containerize complex services or those with many dependencies. Love: a service that’s a single Go binary. A service that only comes as a Docker compose file and wants to run MySQL, Redis, and three other containers? I’m already looking for an alternative.
Ok, I respect your opinion, but remember that there are poor souls like me out there that actually love the opposite.
> Ok, I respect your opinion, but remember that there are poor souls like me out there that actually love the opposite.
All things being equal, why prefer Docker Compose? From my perspective, a single binary strictly dominates any Docker Compose solution.
Isolation from the main system, other services, and other dependencies.
Easy to migrate to new system.
Easy to upgrade and downgrade. When a new version causes an issue just restart it with an older version.
One directory to backup and restore that contains all the services and their settings.
A single binary accomplishes most of those things just fine. Only thing it doesn't get you is isolation, but you can isolate things if you want even then (for example, FreeBSD jails are dead easy to set up).
I use traefik so with docker compose/labels it's real easy to give things a reverse proxy with an auth gate and other middleware
There's lots of solutions, I also find compose hits the sweet spot for integration into my homelab
ok, so you're just using a container with a different name? You can put the single binary in a container and call it a jail if it makes you feel better.
Well yes, but that's the point entirely - you can easily put a binary into a docker container, or a jail, or on a VM. But when something is shipped as 5 different docker containers it is MUCH harder to extract it into something you can run without docker or in a different configuration (like sharing a postgres instance between multiple applications).
One distribution method is flexible and can be deployed however you want, the other prescribes solutions and makes any customization difficult.
I'm confused. The original post said "why docker compose? a single binary dominates any docker compose", (I assumed dominates here means wins over because it wouldn't make sense otherwise?) someone replied giving the additional benefit of containerization, then someone replied to that saying you get all those benefit from single binaries, and if you want the system isolation you can use a freebsd jail, which is like saying "if you want the benefit of containers you can use containers"
Most applications do ship in a single container. whether it's a single binary inside, or a bunch of python or js files or shell scripts is a separate issue. If there is a dockerfile then you know how to build the container. a single binary is easier to build the container for yourself, assuming you also know the binary dependencies of that single binary. Like maybe ffmpeg or yt-dl cli on the path?
Applications that ship as a list of containers are a completely different problem. Those applications are either a victim of their own success, or have developers that come from a particular background.
I've never felt hindered by docker. It's pretty flexible. If I need even more flexibility I'm reaching for k8s, where I am still containerizing any single go binaries
Are you saying it's hard to edit a docker compose file? Being one service or 5 doesn't really make things more complex. I often point servers at my own postgres container instead of spinning up an extra one (so just remove that from the compose for example). That's kind of the entire point of compose, it's so flexible.
I mean I've seen many single go binaries spin up docker before haha
A single config file in a directory beats all.
All my home server apps are in /apps/app-name/ all have a docker config... All data and configurations are mounts under that directory or to my library. It's really straight forward.
I don't have to hunt for different configuration files, each app has what it needs.
I can spin down, move and get back up in a minute or two. All in a consistent way to the platform.
The overhead for Docker isn't near a full vm, it's maybe 2-3% if that.
I don't use Docker but I love reading the compose file when a project has it. It is the quickest way for me to get a high level idea of this project's architecture. It's the best documentation :)
This. Docker compose is so great that you don’t even need to use it, for it to be useful. It can describe so well the dependencies your system needs (alongside proper documentation).
It’s the unified interface for me. Updates, logs, service management, network routing, config and storage mapping are all uniform and I don’t have to wrangle many slightly different implementations of each. Isolation from installed packages is a nice bonus.
I like to have consistent architecture across all my services, and I like to see and configure everything in (mostly) one place. If you're running a single go binary, sure do whatever, but if you're running a go binary, and a *arr stack, and a bookmarking service, home assistant, frigate, etc, it gets out of hand pretty quickly.
You trade a bit of upfront setup (I mean barely now in the age of LLMs) for easier automation and flexibility down the road. I use docker compose to run my local dev stack and qa envs, but then can do whatever I want with the containers for prod. Google Cloud Run, Amazon, whatever. Right now I do just actually use docker compose behind nginx in prod, but I've used basically the same docker compose setup for all my projects that aren't large distributed systems in the last 10 years. And when the system gets too big, I refactor it into k8s, but k8s is overkill for like 99% of my personal projects and like 75% of my professional projects.
for long-running services, I've never taken the time to learn something reliable like systemd. with Compose, I can see everything I need to know in one yml file and the commands to go with it. I also don't have to really understand any dependencies or their own configuration and can cleanly remove everything associated with that service with a few commands.
You can use systemd and docker compose together. A systemd service is a single config file that runs the commands you give it, including `docker-compose` commands. Everything else about your service is still orchestrated via docker, and you can still run commands against the running containers just like before.
A big win with very little effort is systemd starting your docker compose service on boot. You have your systemd config file set `ExecStart` to call your docker-compose command, throw the config file into your systemd folder, and tell systemd to enable it. Docker compose handles the rest.
You don't need systemd to restart your compose service... Just have the service set to restart always or unless stopped.
And if you're running in Windows or Mac it's easier than the OS option usually.
Not OP, but i prefer a single binary on docker, at least i know it won't fuck up my filesystem when i do something stupid with it.
Or at least i can reinstall any time i want, because i know it won't leave any trash in my filesystem and when reinstalling, it won't reuse those outdated files
Its a single config that lets you run multiple services (and can keep them running), upgrades are just editing an image tag.
Don't have to look at cli docs to figure out how to run it.
Doesn't have access to my file system, logs are easy.
I’ve been a die hard native package or static binary proponent for a long time. I still hold that there are real advantages.
As podman has improved and compose stacks have improved, I’ve conceded that containers simplify deployment and updates and isolation, and at this point I find for me the benefits more than outweigh the hassles.
Instead of crafting a set of ansible plays that do all the manual native installation steps, I have a relatively lightweight ansible rig that reads containers and relationships desired from a manifest in inventory, then builds or modifies compose files, then just deploys. Much less futzing. And I’m more likely to keep things up to date regularly as a result
Or, increasingly, dispatch codex to handle it for me
In this way I update home assistant across multiple instances now like magic. I don’t even dread it a little anymore
Some isolation is nice. I say some because I’m told it’s security is not to be trusted, but I still like that even a non nefarious binary has ootb boundaries on what state will persist.
Docker isolation is likely good enough if you keep your kernel up to date and aren't allowing 3rd parties to run code. If you are running a CI service or something you need micro VMs. But to host your nextcloud server, docker is good enough, it's unlikely someone will both exploit nextcloud and the kernel if you are keeping them up to date.
docker isolation is nice just as a management boundary. It's nice to be sure applications don't accidently shit all over each other. Like an application that decides to clear /tmp because reasons, or 2 random ones that both store temp state in a /tmp/state.dat for some odd reason.
My favorite is applications that conflict on ports. So you go to change the port an application is listening on, the docs say PORT env var so you set it, but it doesn't work.
You look through the code (and this is before LLMs) but the config is just a labyrinth of values flowing from god knows where. So you search the repo for PORT and eventually find a PR from 2 weeks ago that changes PORT to ENDPOINT, so you set ENDPOINT, but it still doesn't work.
Then you find an issue that was opened last night that the new `ENDPOINT` value doesn't load from env vars because it conflicts with another ENDPOINT_ prefix on another value, instead you need to create a config.toml and set an `endpoint =` in that, so you do that.
Then you realize that creating an empty config.toml has now overridden all the previously defaulted values into nulls or empty strings, instead you need to run `foo config-defaults > config.toml` and go edit the endpoint one.
Only to realize that there is a secret that the tool uses to encrypt an ephemeral session. if you don't specify one, the tool would auto generate one on startup. The only downside is restarting it logs your client out which isn't a big deal. But now that you switched to config.toml you must set some value there and it can't be empty
Then you're just wondering why the hell did I want to change the port?
I too prefer the Docker compose route. If you give me a single binary, I'll go out of my way to run it inside a compose stack.
I manage my home server with a script that presumes the service is running on a compose stack. So I can run stuff like `stack freshrss restart`. I don't want to touch my scripts, I just add the compose file to some expected path and the script finds it based on naming conventions.
Agreed and everyone’s talking about Docker Compose while I’m putting everything in k8s.
Same here, just give me a compose file and lets go!
A big reason I chose Rust and Sqlite for my project was that end users could just run the small binary and be done with it. https://github.com/sergiotapia/joyscroller
I think you can go very very far with this kind of setup especially if it's just for you.
I am generally not onboard with the “all you need is SQLite” articles that pop up on HN regularly.
But I recently looked into self hosting and several things each wanted their own Postgres container. Immich wants a Postgres and a redis instance.
If you’re building apps targeted at self hosters, SQLite is really all you need.
Immich uses postgres because it relies on VectorChord for smart search.
It's what powers the ability to search your photos for "a dog on a beach".
It's also a fairly sensible decision.
I would rather keep things on postgres but everything wanting its own postgres container is really annoying. Postgres already has support for multiple databases, users, and access controls. I would personally prefer to just set up one postgres instance with proper backups and have everything run against that, unfortunately some self-hosting projects have fairly poor support for that.
You could sometimes get there. The problem is in an effort to simplify the getting-started with postgres (or other databases) most projects assume they would also automate the management of that postgres instance.
The getting started would be a lot different if it first instructed you to install postgres, run a script or create a database and users and configure access controls. Also certain things become easier if you assume you can also manage the postgres instance, like on an upgrade, maybe you just create a completely new database and drop the old one, or isolate tenants within one application into different databases, etc.
In general, I agree with OP that sqlite is all you need for selfhosted applications (most of the time at least)
That sounds like a mess. What do you do when the open source projects you're running need different versions of postgres?
That's a large part of why I lean more and more into just preferring sqlite for applications.
Even in work environments for apps that likely won't have thousands of simultaneous users. Modern SSD drives are so fast that so many apps don't need the advanced features of a full blown rdbms. And I think options like Turso and Cloudflare D1 fill the in between as a hosted option.
I do wish that sqlite had a couple more native types, it works really well as it stands.
I prefer a separate DB instance for each app. Each app is on a separate ZFS filesystem so I can snapshot/backup the app data and DB together as one.
A possible restore only affects that app. Plus I can guarantee the files and DB were snapshotted at the same moment this way.
It may just be a way for me to avoid managing the DB.
> several things each wanted their own Postgres container. Immich wants a Postgres and a redis instance.
Those things are dirt cheap. A Postgres container on my machine uses 200MB of memory. Redis is nearly free. It's a small price to pay to not have services cross paths with each other and mucking everything up.
SQLite has a bunch of cool stuff happening around it right now too like Turso and Litestream.
How do backups work with SQLite? Postgres is nice in that it can run them live without any corruption issues for writes in the middle of the backup.
https://litestream.io/
My experience with this is that all of the projects that are worthwhile, continue to grow and then at some point some kind of bottleneck is hit. Migrations become slow or overly complicated, lookups start taking forever, etc and the app just starts to slowly become unusable the more you try to use it.
You end up in a situation where you are following some complex guide (hopefully written by the actual developer) where you make a huge backup, have to tear everything down, and then hope the migration from SQLite or MariaDB doesn't hit any issues. Postgres is heavier but at the scales of self hosting it should not matter on any modern system. Multiple docker containers even on a Pi isn't something you have to worry about. Solutions like Immich are usually run on a NAS where there would never be an issue
Also you don't always know who your users are gonna be. You could create a cool personal photo backup app for your <20k library, then get people with a dozen terabytes of photos wondering why is it so slow?
Good article. I have many of the same services in my setup.
Self-hosting email fully is tedious/unnecessary, but what you can do is mirror your various inboxes (Gmail, iCloud, proton, non-corporate outlook) with mbsync and use roundcube or some other frontend to view emails and compose.
I use this to easily access mail from different services, and I can also send from any of them.
I also have a seperate service that watches the mail directory and sends a ntfy.sh notification for new emails.
It’s a bit complicated cause theres so many services but here are the quadlets: https://github.com/ebrahim37/infra-template/tree/main/homela...
I implemented this about a month ago because the iphone Mail app was increasingly buggy.
> Self-hosting email fully is tedious/unnecessary
Certainly not unnecessary if you care about privacy and Google not reading your mail.
Is it tedious? Not really for someone with a basic command of Linux and computer networking. Even in the pre-AI days, it only took me an afternoon to set up Postfix/Dovecot, and a few more days to refine various aspects of the setup and to write a bash script that can rebuild the whole thing on a new machine. There is basically no ongoing maintenance at all. The Dovecot syntax changed in Debian 13, and Claude translated the old config file in less than a minute. If you are willing to use a Dockerized solution, it's even less work, but I personally like to run this kind of thing bare metal and control every aspect of the setup.
> Certainly not unnecessary if you care about privacy and Google not reading your mail.
Fastmail exists
"Self-hosting email fully is tedious/unnecessary"
If you IT, why do you particularly consider self hosting email as tedious?
I self-host e-mail, because I am stubborn - I have done it for 20 years. I would not recommend anyone do it if they aren't already, or if they well-and-truly want to be on call for their homelab.
So as not to reiterate, here's an HN comment where I described the most recent trial I endured with it: https://news.ycombinator.com/item?id=46297991
In the last 20 years, I've also run into problems like:
* My brand-new ISP wouldn't allow TCP port 25 inbound, and wouldn't relent. I had to go back to the old ISP.
* SPF became mandatory at Gmail, but I didn't get the memo, and Google started bouncing my mail one random day.
* My residential IP ended up on a "policy blacklist", one it could never get off because it's blacklisted by virtue of being a residential IP, not that it did something bad.
I've never tried to set it up even though I know it can easily be done with things like mailcow.
The downside is never really knowing if your email reached your recipient (maybe it got blocked by Gmail because of your IP, maybe some other reason). If I use Gmail/proton etc, then I can be pretty sure that the other person got my email.
You'll know from bounced email and logs, which you should have monitoring setup on.
It's not trivial to run your own email stack, but it's not hard either. It's more tedious than anything. Once it's up and running, it's low-maintenance. And it's fairly trivial to avoid spam. I get vastly more spam in my Gmail than I do on my self-hosted domains.
The biggest challenge with self-hosted email is getting an IP address that's not blocked or carrying low reputation. Most VPS IPs from e.g. Digital Ocean, Linode, Hetzner, AWS, etc. are going to get blocked from Google (the strictest) and Yahoo, Hotmail/Live/MS, etc. Cloud providers often block incoming and outgoing mail ports, too. I have a Digital Ocean VPS and my account is old enough to be grandfathered into the old policy, which means my email ports aren't blocked. I used to have no trouble sending email anywhere from my VPS, until I downsized it and got stuck with an IP in a bad reputation CIDR. Now I use smtp2go for relaying outbound mail, but still handle all inbound mail. 10/10 score on mail-tester.com, and I never have any problems or need to do any maintenance.
A few major vendors (I want to say Microsoft was among them) were infamous for silently dropping email without bounce backs. They'd perform then whole protocol sequence as if they accepted the email, then simply never actually put it in that user's inbox or even spam folder.
People should read the parent post and stop spreading FUD about email. The only part that isn't easy is the IP requirement. If you don't have a fixed public IP, preferably v4, then self-hosting email is out of reach, but so is hosting other things like game servers. It's the main cost of self-hosting, and one that isn't big relative to all the benefits. Privacy is the biggest one I think, and so is the ability to create unlimited aliases, one for each company/website.
Because it's not about setting it up, it's about not getting outgoing mail perceived as spam by someone. Borderline impossible without scale. Even doing everything correct is not enough.
and yet I manage to do it for multiple domains, for over 20 years in the UK.
You can do this with Proton?? How does it handle decryption/encryption, etc?
Would absolutely love to do this, proton frontend annoys me at times.
I actually don't use protonmail but you should be able to access your mail with IMAP (both iCloud and Gmail support it so proton should as well).
If by encryption you mean PGP etc, that's handled in roundcube I believe.
mbsync mirrors your protonmail inbox to disk, dovecot shares that mail folder through IMAP, and roundcube or some other frontend connects to dovecot.
For gmail/icloud IMAP, you can't use your regular account password, you have to create an "app password" and use that password for mbsync.
Yeah they have integrated end to end encryption, this made me go digging, and I found they have a bridge app that handles it.
https://proton.me/support/bridge-for-linux
Excited to set this up, I had assumed it wasn't a possibility.
> . I do have a 4k TV, but it’s not extremely large, and while I can tell the difference in resolution between SD and HD, for almost everything, 720p seems to be good enough for me.
I am actually jealous of you, I can tell the difference between 4k low bitrate and 4k high bitrate, 1080p and 720p are like mud to my eyes.
I got a VHS player and a load of free tapes. It hasn't changed since I was young (well they've degraded a little), so if I enjoyed it then I can enjoy it as well now. In some things I try to resist a little the encroachment of higher standards.
But not in everything. I can't drink a cheap scotch.
The difference between a 4K bluray encode by pirates who care, and a 4K WEB-DL from Netflix and co, is simply staggering.
Running all these services without a declarative/standard way of deployment, ie docker compose or k3s/k8s is a bit mind boggling to me.
I self host on a bunch of different hardware, OSes, providers. I want the least fiddling as possible.
It's been really interesting seeing the divide in the comments! I have a completely different philosophy from the OP on how to self host haha, and it comes from not wanting to fiddle with my server all the time! But they say our ci/cd is fiddly! The whole point of it is to NOT be fiddly! People gotta start figuring out how things work instead of throwing spaghetti at the wall.
In my experience, when something involves words like "declarative", "deployment", "docker", "k8s", that's a sign that it will actually involve more fiddling rather than less. But that's just me.
I'm running like 10 services on a VPS. I'm a professional software engineer. Do you think I want to ssh into my server and find a bunch of single binaries all over the place to run them? What if a service stops, how do you automatically restart it? How would you automate creation of a qa environment when you create a PR on github so that you can test your changes before they go to prod? Gonna ssh into the server and pull down your repo and build the app and run it?
It takes like a day to figure out docker compose and like a week to figure out k8s at a level where you can get things done. Honestly, you can get something working now with LLMs and have them help you understand it in probably less time.
Of course I want a declarative deployment for my complex home server. Do you run like 1 app on your home server? It's not "fiddling" if you know how these things actually work.
I imagine it's more fiddly to figure out how to wire up all your individual services to Uptime Kuma and Homepage. You don't even need Uptime Kuma with docker compose, instead you just set a container restart policy and have a healthcheck endpoint.
Noone builds anything without a vision of what it will look like. Blueprints are declarative specs. Why bother with instructions if you dont need to?
Many projects, like Immich, only provide docs for docker images/compose deployment.
I agree. But you could easily automate the systemd/podman based setup the OP has using ansible, if you wanted a declarative approach using OP's tools.
For my part, my home server is a single compose file :shrug:
Do you not consider systemd services to be declarative/standard?
How do you declare the dependencies though? The unit file only covers starting the process.
Pets are OK sometimes! They do need care and feeding, though; some of us don't see that as chores, instead delightful hobbies. I can definitely see the perspective of "there are better things to do", especially if you don't like those things.
Pets are totally fine if you consistency/reliability arent that important. Homelabs can be fun ;)
I mean I'm not self hosting my password manager like this guy, but my self hosted services have better uptime than github actions
28 is a lot of pets.
There is something satisfying in handling a home server as cattle though!
I don't do it because I don't care; I do it because it is super satisfying when it has been boringly running for weeks without me doing anything to keep it up beyond paying the electricity bill, upgrades are extremely predictable, etc.
Boredom is a feature :D
Especially now when you can snap your fingers and get a semi responsible codified version. Heck just do what you want to them ask the LLM to port your work to something marginally better than a totally hand cobbled bucket of bits system.
That’s what I’ve been doing. Managing this stuff by hand is awful and I’m not a sysadmin who is good at it. I just added single sign on to my home network for a wide variety of various services. Some of them didn’t even support SSO but a quick AI patch fixed that. There is no way I would even attempt that by hand let alone succeed.
I saw statements to the effect of “this would be nice but it’s a lot of work” several times in the post. I don’t understand why that has to be a blocker anymore. All this tedium is gone.
I see people say the same things in threads discussing open source agent harness repos that are MIT licensed. I don't understand how it isn't obvious to them at that point haha. I don't think people realize they can just... do whatever they want to
Self hosting is great. The access part is hard.
Pangolin was the missing piece I needed, to not expose my servers to the internet and easily have my services publicly accessible with built in auth.
Tailscale kind of worked, but it was a hassle needing the client on mobile device and it being a VPN.
With vibe coding, I have more of my own apps running than pre-built.
Wireguard is relatively simple. The VPN on my phone automatically connects when I disconnect from my home wifi, then it's like I'm at home.
Pfsense and maybe openwrt allow split horizon DNS that allows Wireguard tunnels to work seamlessly whether you’re on or off your home network. Took me years to figure out but it’s what I use now
Not requiring friends/family to use a client to access my services is so much simpler.
I also use Adguard which act a VPN to filter traffic. I dont feel like messing around to get both working.
Why the downvotes? Genuinely curious, I do this and don’t know of a downside after setup (with help from {AGENT})
What tools did you create yourself?
For people interested in a minimalist setup, Tailscale + Samba can go very far. I personally use Foobar2000 and VLC on the client side.
It’s not pretty nor great with bad connections, but there is basically no maintenance.
This, but plain Wireguard instead of Tailscale. This assumes you have a public IP. But you can't do true self-hosting without one.
foobar's rock solid. I have a portable copy on a mounted NAS share and it's great for transcoding and being an ad hoc music player while setting up new computers
Just a note on OP's hardware choice. I used to have a Beelink with the N100 CPU. That thing overheated and reboot intermittently after couple months of usage. I suspected they overclocked the CPU to look good in benchmarks. I tried to underclock in the BIOS. The temperatures dropped but the setting was reverted after the next power down. Nothing in the BIOS stuck. Their support was worse than useless. The machine ended up as spare parts.
You are better off getting the used business mini-PC from HP/Dell/Leveno/Asus. They have better specs and run great.
This. Or just buy an N100 motherboard (Asrock makes one) and put it in a standard ATX case with a proper cooler, power supply, and enough space for expansion cards and hard drives etc. Maybe you will want a GPU in the future, even if it is unncessary now. Small nonstandard form factors always come back to bite you sooner or later. It's better not to cut corners and build something modular.
Most likely something wrong with the cooler, maybe it's not making proper contact with the CPU. Should be repairable if you give it a try, but yes, the used thin clients are higher quality.
I did clean the thermal paste between the CPU and the heat sink, and re-applied new thermal gel. That didn't help. It was the raised voltage preset in the BIOS. If I lowered the voltage, the temperature stayed low and it would run fine, until the next power down reverting the BIOS changes. The box was too small, the heat sink was too cheap, and the fan was too lacking to be a proper overclocking setup. It would run fine out of the box for couple months until the CPU and hardware worn out due to stress and heat.
Hmm that's too bad. I have bought a few of those (different models) over the years to use as HTPCs for different family TVs. Only one had cooling issues (bad fan bearings). I avoid the extra small ones and just get the NUC-like ones.
New CMOS battery might’ve fixed the BIOS resets
I have the opposite experience. Plex, nextcloud, minecraft servers, and ~10 websites are all hosted on a Beelink N100. It stays cool and quiet.
Just another data point, I got a Beelink S13 Pro N150 running similar things as OP and haven't had any issues. I made changes in BIOS (set "Chipset > PCH-IO > State After G3" to S0 so it boots up again after regaining power) and they are sticking for me.
Same here, N150, no issues. Don't think I've made any changes to the BIOS.
Fans spin a lot near-constantly though, I think it is the on board M.2 SSD that is hot, I need to get around to adding a thermal pad or something one of these days.
Maybe I skimmed your blog too fast, but I'm surprised I didn't see a dns ad blocker. Thats the first thing I'm seting up when I have a more permanent.
I run Adguard on my router, not on my server. I did write a subsequent post about the rest of the stuff I run on my router and my Proxmox cluster, but forgot to mention Adguard.
Dns4eu is free, has adblocking, and you can deploy it right now.
Never tried it myself but I’d be afraid of stuff being randomly broken. Eg at some point pi-hole made YouTube history not work.
To each their own but running that much stuff without proxmox would drive me insane. Especially the native stuff straight on the machine.
With some small systemd configs each native app is essentially a container these days. The overhead of unique kernels per app would really restrict what could fit on a small box like this
DynamicUser and ProtectSystem keep me sane.
Like the other person said, proxmox allows containers or full VMs, and the base OS takes less than 2 GB RAM.
My jellyfin LXC idles at a few hundred megs of ram, shares the base debian kernel, and if anyone does ever break in, they don't get access to anything else, anywhere else on my server. (All media is also mounted read-only to the jellyfin LXC.)
I really, really like it, just set it up recently. Everything in its right place.
Much prefer LXC + IaC to rebuild them, but much of a sameness
Heya, nice read. I’m wondering how you managed to get LetsEncrypt certs minted when your caddy isn’t exposed to the public internet. Doesn’t LetsEncrypt need to make a request back as part of the handshake to mint/refresh the certificate?
Would love to know as I wanted to do the same in the past and ended up self signing certs.
Use the DNS-01 challenge.
Much appreciated
As someone that loves docker compose and doesn't mind services with the benefits of software better than sqlite, I guess I'm the minority here. I wish more apps were based on postgres. I also have never had a problem with my publicly exposed applications. Not once in 8 years.
Nah I thought their setup was pretty ridiculous and seemed hard to maintain. I run all my stuff via docker compose and am very happy with it. I don't understand the comments praising single go binaries, which I've actually seen reach out to the docker engine api and start containers for their dependencies
As long as you can use one instance of Postgresql. There's another comment thread here[1] about each thing wanting its own Postgresql container.
[1] https://news.ycombinator.com/item?id=50111944
8 years to now may not be indicative of what's to come.
People with script-kiddie level skills now have access to nation-state level attack expertise via LLM.
I'm a big fan of Tailscale
But can you still call it "self-hosting" when a corporation is now controlling access to your own network?
I mean I'm just lazy. I could use Headscale or implement my own wireguard control plane.
The apps and data are still all on my hardware, Tailscale just brokers the wireguard connections and can't see the traffic.
I get what you mean though. It's a link in the chain for sure. Never had an issue with it so far though.
Alternative for music:
owntone + shairport-sync
Owntone holds your collection on the sever and can stream to airport, airport2, dlna.
Shairport-sync turns a raspberry pi zero 2w, or orange pi zero 2w into an airport {1,2} speaker. Or use a (super cheap) esp32 airport 2 from aliexpress.
Can sync every room in the house with owntone OR "I want my phone or laptop's sound to come out of the speakers in this room"
Owntone has access to your paid spotify account if you want that. Can stream to your phone with http (remotely with wireguard vpn). Also has mpd controls if you want those and can use apple remote app on your phone if you like that.
Owntone, pihole and wireguard is reason enough alone to put an old laptop into service as a home server.
With the recent change in Bitwarden's license, not sure for how long they will continue to be self hosting friendly.
I am also using Vikunja instead of trello.
Tbh in general I am fine if the self hosted app uses postgres or a container Db instead of sqlite only.
I have gone down the self-hosted rabbit hole, like many it all started with setting up a pihole on a raspberry pi.
I have since gone down the jellyfin + arr stack route. Resource wise the entire stack uses very little. I have Sonarr, Radarr, Prowlarr, Bazarr, FlareSolverr, Jellyseerr, SABnzbd, qBittorrent, Gluetun, qui, Decluttarr, Recyclarr, arr-dashboard all running on my dell optiplex (previously on the raspberry pi).
All up this uses about 2GB of RAM and 0.5% CPU.
I found it extremely fiddly to set up, understand what each component does and have everything integrated well across all my devices.
Without the help of an AI assistant I just wouldn't have had the time to debug every little issue. It took me a good 12 months of tinkering around work hours to have everything stable. However now it runs along fine with very little intervention and I'm able to share it with friends and family.
So the resource argument I don't think is valid. However the fiddly mess I do agree with. Although personally I found it fun and rewarding when things finally worked. But I can understand why people would not want do spend all that time on it.
I'm puzzled that OP seems stumped on syncing keepass when they're already using syncthing. I've been using keepass+syncthing for years, and I've never had any issues.
I considered KeePass+ Synching but tended to run into the same issues as with Owntracks - Android is hostile to always on background apps. Then when it came time to onboard my spouse, KeePass wasn't an option.
I'd be interested in a list of services he uses that he could self-host but chooses not to. I think the only one he listed was email.
I recently bought a Lenovo Mini PC (01Q8X10) for under £300 that has a Snapdragon X1 CPU (and 32GB LPDDR5 RAM!) and it's turning out to be perfect for self-hosting running the newly released WSL Containers on Windows 11 with WSL 3 (it doesn't fully support Linux yet). Podman had some issues with WSL 3 that they're still releasing fixes for, but it's all working with that too.
I don't know why so many people host personal forgejo / gitea instances. You only really ever need a server with ssh access (and cgit if you really want a web interface).
Just FYI you can get a M1 Macbook Air with a broken or cracked screen for about $250 on eBay. It's a headless device at that point but it has a keyboard and touchpad.
I use it for my Home Assistant setup and more. It still has headroom. No fan, very low power consumption. Not true Linux container stuff but pretty much everything works on it in this area.
My go-to for low power consumption is a Chromebook running some vanilla flavor of Linux. I have some HP units that will run for almost a day on battery with the lid closed. $25<$250.
How do they do with staying on and then restarting everything after power off? Can they automatically repower?
I don't have my embedded macbook fully configured but gemini says I can get it to repower itself.
By repower I mean if the power goes out and then returns the machine needs to power on and boot.
I have a couple of older Chromebooks I use in rando locations (basement, shop, car, rv, boat etc). And they are great for that casual scene-specific use. Very much recommend and amazing what you can get for sub $50 easily.
One can create a gmail account like my-robots-73489@gmail.com to sync everything too without touching your main google account. Very handy.
[edit: one thing about the M1 is that it has native h.264 h.265 and prores hardware encoder decoders if you are doing video intake of multiple streams]
yeah, they make a great homelab, especially with a fast external drive or 4
Great write up, thanks for sharing. I run some of the services mentioned, but got some good ideas from the article and the comments.
One suggestion in place of MPD is lyron (https://lyrion.org/). I’ve used MPD for years but recently switched to lyrion and for me, it’s a lot easier and suitable when you have many clients. It’s nice to have a single (nice) UX/UI for my music and all the “speakers” on the local network.
I've been building and self-hosting my own analytics platform[0] for 13+ years and never had any major issues with it.
The advantage is that if I do want a specific feature, I can just add it, and I am never worried about subscription limits or surprise charged, because I host it on a fixed-cost VPS.
Things don't really need maintenance and don't really break by themselves, the only time when they can sometimes break is when changing things.
You do need to have a proper setup though, for stuff like disk space alerts, knowing the common gotchas of disabling settings that can cause issues, etc.
[0]: https://uxwizz.com
Is putting your reverse proxy in a VM not secure enough? What can break out of a sandboxed VMM?
I suppose there was a KVM exploit on twitter a while back...
/Aside/
Realizing (not just now, btw) that some folks have such a complex setup in their home networks... Make me go :wide_eyed.gif:
I dont even know half (most?!) of the apps/services listed there, let alone know why one would use, or even self-host, them..
/Aside
> I need to think about realistic ways to handle potential server failure without compromising the functionality of my home
Sounds like his family is in a pickle if he’s out of town and the server goes down.
They might have to turn on/off some lights using the buttons on the smart switch, yes. Or maybe lose access to Jellyfin for a while. But on whole things have been pretty reliable.
How do people use let’s encrypt on a host that is not accessible from public Internet? Do they manually renew every three months through DNS challenge?
I used to have my server publicly facing, but now I use Caddy with automatic DNS challenges.
You can automatically renew through DNS challenge?
Yes.
to each their own i guess. i love docker and exposing my services publicly. and fortunately i bought 64gb 3 years ago :)
For the self hosters: docker containers or not?
They are so easy / portable but perhaps systemd is a better option as the author suggests.
Podman + systemd is how I run production workloads at work.
For us it's the perfect balance between simplicity, observability (Beszel) and reliability. Not the right tool if you have dynamic workloads but definitely the right tool for boring and predictable workloads.
At home I run K3S, it's awesome. It also breaks a lot and does add a whole bunch of complexity that I would only recommend if you enjoy tinkering and fixing things.
I second this, it's great. Commonly called "Quadlets".
I also use BGP advertised pod networks so each pod gets it's own IP address that's directly routable (l3 ipvlan) with the rest of my network (hint, don't use 100.64.0.0/10 for that if you use tailscale), so I don't bother with NAT on the host itself or port mapping or port conflicts. It also makes caddy in front of it super easy, barely an inconvenience.
My server is running on NixOS and I went with systemd containers - really happy with my setup. It was not at all difficult to set it up with Claude
I have a stupid question: if I'm using nixos why do I need docker or systemd containers? Especially with systemd sandboxing
I prefer docker containers (on proxmox VMs), but that could just be the familiarity of a few years experience talking. It does add some distance from 'bare metal' but it also makes backup/restore, and therefore portability, quite simple. I backup both the VMs and the 'persistent' container data that I always create dedicated volumes for, so I have a couple of different avenues and granularity for restoring.
There's nothing else out there, that I know of, that offers improvements on any kind of scale that it would be worth my migrating to.
I do hear good things about podman, so if I was starting from scratch I'd look into that. I have one podman instance running UnifiOS, but that's only because there's no official docker compose from Ubiquiti (and maybe that's telling in itself). Podman feels more opaque to me than docker, but I have years under docker and essentially zero under podman.
I had an easier time moving a project from docker compose to k3s after wasting so much time with podman compose (maybe expected considering than k3s is still root daemon)
I personally use NixOS and a mix of systemd services and Quadlet-Nix for the few containers that have too many dependencies and moving parts. Works well.
I had not heard of quadlet-nix. How interesting and tidy!
Currently testing out FreeBSD jails, but with OCI images and Podman :)
https://daemonless.io
BastilleBSD is another great way to run FreeBSD+jails using their templates
https://bastillebsd.org/projects/bastillebsd/
https://github.com/BastilleBSD/templates
Container manager on Synology. Don't have much choice. Some stuff can run native but it's easier to run containers.
The big issue with Synology is the really crappy db logs driver they use. If a container has been running for a while it takes an age to see the logs via the Web UI.
There is a solution though: https://github.com/telnetdoogie-labs/synology-docker
There are really easy to user docker app managers out there. The author says 'I already use systemd'. I mean, I guess I'm using systemd, but really I'm using e.g. runtipi with an app store, web-based config and logging, and a super easy app template. I can point claude at any project with a docker release and in a couple minutes have it show up in my app store to install, run, simple data backups, etc.
For me, I say it depends on the service. I use Coolify for docker apps usually smaller apps, and LXC when docker won't cut it for whatever reason.
Personally I use docker with portainer to handle the stacks/containers, have done for many years. I like it.
Yes. NixOS running Podman containers is where I ended up after trying almost everything :)
Have you tried quadlet-nix? It sounds like it would fit your setup pretty nicely :)
I have considered it & heard good things about quadlets, but since I am more of a Compose person, I use compose2nix instead (self-plug).
Docker containers are fine. I've moved between 5 NAS' over the years and docker has enabled it to be seemless.
rsync data directories. docker compose up. Done.
On the consumer or even prosumer nas products which have a semi-locked down system (ie Truenas/qnap/Synology). Technically you have root but as soon as you update the OS all/most of the changes end up being broken. Yes you could probably do something and deal with it. But personally I've spent too long babysitting it on the first and second nas that I no longer care.
Yes, especially if you running a multi-node setup. I'm still running on HashiCorp Nomad.
Why not have your cake and eat it too with quadlets?
Quadlets rule, I have tons of them configured for my professional and hobby/self-hosting services. Super convenient, I wish there was something equivalent for my Mac (launchd files just don't cut it).
For several years, Docker with portainer
Most of the time projects provide a ready to use compose file. I use dockge that provides a simple web interface for compose configs, so just paste in the compose file there and run it.
I forgot all about MPD. MPD is great, I used it long ago, just exposed to the public internet. Nothing bad ever happened.
I am surprised author said *arr stack is a mess, it's probably one of the few things I haven't touched much since setting it up and it saves me a ton of time managing new media.
Seems like dealing with this is a part time job. I commend it though.
I always love reading other peoples self hosted stacks. I never heared of SnapCast but it sounds interesting! Will check it out.
People twist themselves into a knot talking about AI as the future. Its more of a bubble waiting to pop. Self hosting is the real disruptive future.
The hardware to accomplish most things that we rely on from cloud vendors is absolutely trivial. My own home server is a used super cheap and tiny machine. The hard disks don't even fit in the box and yet it has plenty of CPU and memory to spare.
The software is getting easier to configure year after year eventually creeping to the point where elderly people too lost to use a remote will be running this stuff. The real challenge will be solving for easy access to private networks from public spaces across the internet. Right now everybody seems to be centering on vendor solutions like TailScale for this.
I self host (I have a large overlap with the services in TFA), and love it. I will never go back and think that more people should self host (or at least...join small groups of people who designate one person to manage it for a small number of people).
And believing that the current AI economy might have a period of rapid devaluation in the near future is pretty reasonable. But, much like the original .com bubble, the idea that AI as a technology will not survive and thrive through such a pop, and become an ever more important part of our future, is a belief that I can't comprehend. Anthropic and OpenAI very well may go belly up. AI as a technology isn't going anywhere.
And as much as I'd like to believe that more people may self host more things in the future, self hosting will never, even if it was adopted by literally everyone, become as big a deal as AI, even if AI hit a wall tomorrow and didn't get any better than the current frontier models.
Self hosting has never been easier and I imagine the average person has never been further from trying it. No one will ever get anywhere close to hosting this much sruff
Self hosting is mostly a hobby like home brewing beer or woodworking. It’s more accessible than ever but there’s still little incentive for the average person to do it.
The average user is much safer putting their photos on iCloud than a home server unless they are tech inclined and personally invested in keeping it secure and backed up.
> Self hosting is the real disruptive future.
Self-hoster reporting in.
I got into it about a year ago. Bought a refurbished old Dell mini PC for 300 bucks and turned it into a home server. Debian + several docker containers.
I don't know if this is the future, but it feels disruptive. I think it is amazing that I can run my private home streaming service alongside a bunch of other cool things on old hardware and the experience is in many ways superior to actual services that I am charged to use.
> Jellyfin
I wanted to self-host Jellyfin, but hit the problem. It's bad at the main thing that player should offer - media playback.
If you open https://jellyfin.org/docs/general/clients/codec-support/ you can see that this doesn't support HDR and that doesn't support H265. There is Jellyfin Desktop and it breaks when display scaling is over 100% (what year is it?). Some clients have atrocious subtitle support. And then none of them apparently can't properly pace frames so I still see judder as if I'm on 60 Hz screen (24->60 can't be even translation). But my screen can do up to 165 Hz, so that should never be the issue.
In then I just went back to playing media files from Samba share with MPC + madVR. No automation or catalogization or anything, but it is supreme at actually displaying things, and that's what it's important to me.
Fair, I just never use the web clients. I mainly use the Android phone or TV apps, which do support H.265. The Android TV subtitle support seemed ok, I think I did have issues with the Roku app and subtitles.
I understand use cases for Jellyfin and selfhost it myself for music (in combination with incredible Finamp[1]), but for movies/shows plain MiniDLNA[2] works like a charm and is accessible from any device.
[1]: https://github.com/finamp-app/finamp
[2]: https://minidlna.com/
I used jellyfin-mpv-shim. You select what to play from any ui and it starts playing in mpv, which seems to have excellent support for everything you would want a player to do.
Tried it too, it still had judder as all other clients I've seen. I think it's either server software problem or maybe all clients fail to do this properly.
After Google Assistant died I immediately installed Home Assistant on my old PC. Not as good, but at least it doesn't take 5 seconds to turn on a light like Gemini
I stared self hosting in 2020 when Raspberry Pis were super cheap (not anymore :( ) and lots of things started becoming available in Docker. Haven't ever considered going back to the cloud. I love taking back full control.
Same, Covid was a great time to spend shifting things from clouds to home. I have a small fleet of Pi4's and 1 newer 5 (for running Omada locally). Cost has shifted from subscriptions ($$$) to maintenance (time) :)
I highly recommend https://github.com/mother-of-all-self-hosting/mash-playbook for those interested in secure containerized self hosting.
A tremendous number of services are supported
non-root user execution, read-only, and cap-drop where possible
i used to setup a NAS. it was closed few years ago. plan to buy a new one. you are a legend.
Out of curiosity, do you have any concerns about backups (especially for e.g. VaultWarden and Immich)? I've been pondering running them, but figure that I'd need to back up to The Cloud for me to feel less concerned about losing everything, at which point: what's the point? Curious what you're doing on that front!
I use both Restic and Syncthing. I have considered putting encrypted backups on a cloud blob storage, but my preference would be to set up an offsite backup at a relative's house.
I use restic to create encrypted snapshots on external cloud storage for the data I care about. My media is stored on a nas which covers me for drive failures. If I accidentally delete everything or my house catches fire I can download my personal data again and just redownload the terabytes of media.
About to retire the NUC I used for 10 years running arr stack, jellyfin, home assistant, pi hole and a few more containers. Bought a new bee link mini pc, new Intel CPU, 500gb SSD and 16gb ram for less than $500.
The best thing to self host (if you have some international family/friends you trust) is a VPN. Get each household a raspberry pi joined into one vpn. It means you actually get local tv to work (unlike Nord vpn etc). Each pi sits on a guest network so traffic passing through can't touch anything in the house
I have been self hosting for over a decade and recently have seen more people in my circle doing that and I have been helping them.
At first it was hard but then i made it into a product. I am working on Tarvis.io which makes self hosting accessible to non technical users as well. It comes with installation assistant, ssl domains and backups inbuilt into it.
You can install open source apps from a catalog or drop any github repo and it installs and run it for you.
With Tarvis, I have been self hosting more since I am able to install and delete apps with one click, it makes it easy to try things.