Just got into this business and I have a few ideas from running a personal wiki to Bitwarden, Plex etc.
I'm curious what you folks are running so we can all share and explore more ideas.
Thanks!
Maybe not what you expect for an answer, but I'm self-hosting a telegram bot that does a bit of scrapping of a certain site I'm interested in, and serving those messages to a couple of subscribers (around ~10 people).
I know, it is a very low user count, the interesting bit is, I wrote this bot in Rust using async with Tokio, and it's running on a Raspberry Pi 2011.12 with 256 megs of ram and a single ARMv6 core.
I was able to compile a fully modern language for a 10 years old architecture, and on avg has a load of 0.2 of CPU usage with 10 megs of ram!
It was a very painless process using cross-rs, and I'm still quite surprised how well has worked until this day.
>>> but I'm self-hosting a telegram bot that does a bit of scrapping of a certain site I'm interested in, and serving those messages to a couple of subscribers (around ~10 people).
Very interesting, I can see how this can be more practical than you think
I used to run a lot at home. But i realized that "support" stops when i would die.
My family, who is a- technical, would have not even have a clue on where would be what (since they freak already out if an icon on their phone is move 1 millimeter). So from that nonfunctional requirement for my own home solutions, I decided that self-hosting would not be the best choice, since it would be too dependent on me.
My family's knowledge stops somewhere at the concept of that pressing the B makes a word bold in Microsoft Word. Trying to explain how to run a script on a certain OS via a certain connection would be abracadabra.
The only thing i can imagine is if there would be some kind of paid service that would mirror your own home solutions and provide support for the long run if you fall away and which offers, when there would be again someone more technical to again transfer it to a home environment. (they would take care of new versions, bugfixes, change requests, databackups and so on).
Something like that, but that does not exist (well... as far I know).
So therefore, i try to host every solution at the places which are the simplest to understand and document what needs to happen if I would fall away. I also try to minimize the number of services. So I use office365 and onenote to document everything. (This used to be on my own hosted wiki).
Based on the nonfunctional requirement of maintainability.
I think there is however a need for a service that offers something like the above, to provide long term support for selfhosted environments taking into account all kinds of standards. That would possibly enable the self-hosted direction again.
This is a great thread. I have had a home server for almost 10 years at this point for files & media, and added a second dedicated dev server this year. They both run FreeBSD and I honestly couldn't imagine my life without them. I recommend to people all the time that they get a home server, even if its just a used NUC, but not too many people do for whatever reason.
My file / media server has roughly 20TB online and has movies, games, books, personal files, etc served with NFS & Samba. I don't use anything fancy like Plex or Jellyfin, just XBMC which works fine. It also runs the usenet stack nzbget, nzbhydra, etc and has a virtual machine running transmission through a VPN. I also have a custom ruby script that transcodes files to x265.
Once you get into the home server biz, the most important thing is to have a great case. Mine is the Fractal Node 804, which IMO is truly awesome - a big cube that holds a microatx board and 8 drives, and its actually pretty affordable. For hardware specs, I started dirt cheap and then upgraded too much, building a 12-core / 64GB setup that sat at 10% CPU utilization most of the time. I've since downgraded to a i5-10600 w/32GB RAM which is more than capable.
This is beyond the scope of the question, so I won't get into details, but I would also encourage any developer to have a dedicated dev server as it helps you to better understand performance and deployment.
- Jellyfin (like having a personal Netflix, comes with a built-in web interface and has clients on Roku, AndroidTV, and AppleTV, among others—I spent north of a decade fucking around with XBMC/Kodi before settling on this, way more watching stuff, way less fiddling with the system, vastly better than Kodi for my purposes, anyway, far more usable by the non-geeks in my house)
- A Minecraft server for the kids. [EDIT] Buying actual Minecraft is such a huge pain in the ass now that I did try to convert my kids to Minetest with a Minecraft-alike plugin, but they rightly called it out as both very different-feeling and worse. I felt the same way when I tried it but hoped they'd be OK with it, but no, rejected within a minute and it was hard to argue with their verdict when I'd come to the same conclusion myself.
- Pihole. (ad blocking, local-network DNS which is handy sometimes)
- Samba (alternative works-almost-everywhere access to all those files Jellyfin is serving)
- Prowlarr (torrent meta-search, lets you search across dozens of torrent search engines with one query, very handy)
All of these are hosted in Docker on an old Lenovo workstation I bought off Ebay, then added extra ECC ram and a bunch of hard drives to. Linux, but the bulk-storage drives use ZFS.
I'm considering adding:
- Nitter (much-better Twitter interface—though IDK how long this is going to be relevant, the way things are goin)
- Some kind of solution for hosting YouTube stuff. Jellyfin can kinda do it but it's far from ideal. And the 3rd party plugin for it breaks some Jellyfin clients, so that's useless, plus its mode of operation isn't ideal even if it did work, so it only really works OK for YouTube series that happen to have TVDB entries so you can treat them as TV shows. Ideally this would let me use admin-level accounts to browse actual YouTube and save videos from it, then make only saved videos available to non-admins (with the ability to filter some categories or channels out, as you can with "libraries" in Jellyfin, so I can save stuff the kids shouldn't be seeing yet, without their gaining access to it). I dunno what this will end up being, yet, if anything.
- Maybe something for music that's better at hosting music than Jellyfin is. It's OK but could be better.
I tried to use Jellyfin twice, and my problem every time was that in comparison to Kodi it produced very broken results when scanning my library. E.g. wrong show names, almost arbitrary movie names, etc. Have you faced that?
What is your take on Plex vs Jellyfin?
I am considering the move, but haven't really had a lot of problems with Plex other than it not being open-source
Never tried it. From reading others' comparisons of them in the past, I think the main material difference is that Plex has clients on a few more-obscure platforms than Jellyfin does (mostly TV operating systems other than Roku or AndroidTV). Me, my server's not strong enough to live-transcode past maybe SD resolutions, and a bunch of my stuff's in h.265, so I'm stuck buying nicer streaming boxes that have hardware h.265 decoding, rather than relying on e.g. Samsung's built-in OS (I eventually even had to give up on Roku, when I got enough h.265 content) so it doesn't matter a bit for me.
Plex just works on LG TVs (i.e. has a native app), Jellyfin doesn't have a native app. Plex is also "good enough" for most purposes and actually relatively painless to set up. It gets shit on a lot here but I'm grateful it exists.
Nextcloud (files, contacts, calendar, GPodder sync), Home Assistant (and Node Red for automations, Mosquitto MQTT server, ZWave JS), Joplin Server (notes sync), Plex, Vaultwarden, Miniflux (RSS server), Pihole, Syncthing, Linkding (minimal bookmarks with tags, description, etc), wireguard server, Valheim game server, ntfy (push notifications for server alerts and makes a nice websocket/SSE backend for some simple self-made tools), some privacy frontends (libreddit, invidious, nitter).
I have both MariaDB and Postgres databases running and Caddy as my reverse proxy.
I am currently using Obsidian and thinking of moving back to Joplin. Is Joplin now able to sync changes when detected? or is it still limited to periodic sync?
Home Assistant, Plex, Freeswitch, backups with borg and rsync, various blockchain nodes, IPFS, Whisper ASR via API, Stable Diffusion with web interface, Gitlab, Appsmith, tons of storage via minio and probably a bunch of other things I’m forgetting.
I've been hosting the Sonarr/Radarr/Lidarr/Prowlarr stack, Plex, HomeAssistant and a Misskey server and a Matrix server for close friends. I had considered Bitwarden too, but at $10/year I'd rather just pay to have it hosted reliably, while I self-host things that don't outright make it hard for me to do stuff if they happen to go down.
I just was introduced to Tdarr the other day, and it is a great addition to the *rr stack. It does a great job in optimizing media files based on what you value (size, bitrate, container, etc)
I'm aiming for moderate bitrate, 1080p, hvec containers. In running through 900 of the 5000 media files I've already saved 3.4TB of storage space.
FWIW, BitWarden syncs your passwords locally. The only time you would run into issues is if you need to access using a new device, or a new password that you haven't synced (ex signed up today but haven't used the BitWarden app).
The only scenario self hosting BitWarden is a problem is if I am away, there has been a power outage at my house and nobody I can ask to switch on the server, and I have misplaced my phone and laptop and need to access my passwords. Pretty unlikely. I have some neighbours I can trust too, so I'm pretty sure that scenario will basically never happen.
(I will need to make sure everything starts up properly on startup though. Or maybe my 12 year old neighbour will get a lesson in linux.)
My SATA controller died once in my server so I got to test my disaster recovery (I backup databases and some Docker volumes nightly offsite with Duplicati). Got Vaultwarden and Mariadb back up on a cheap VPS from a database dump in no time.
I hadn't realized that Bitwarden would work fine locally (in hindsight it makes sense as otherwise it'd be useless without an internet connection). Maybe I'll try self hosting it then.
Keep in mind that Bitwarden also works offline. If ever your client cannot reach the server, it will still have all entries saved locally. You can then back up all that and restore later, if your server should ever have a problem
I run a Synology NAS and I use their photo backup and management software. I also run Monica, Wallabag, PiHole, Borg Backup, HomeAssistant, Firefly III, BudiBase, Tube Archivist.
I finally got around to replacing my personal hosted WordPress site with a VPS, allowing me to get started with self-hosting apps (which is something I'd been planning to do for years).
I currently have a Hugo blog, repo hosting with Gitea, a self-developed travel microblog/tracker. I'm thinking about playing around with Mastodon and Matrix at some point, or perhaps Solid.
The last thing I really want is a cataloguing tool for tracking books I read/films + TV I watch/games I play/music I listen to. This would replace the various social cataloguing sites I currently use (Goodreads, Letterboxd, Followmy.tv, Grouvee and Last.fm, respectively).
I've not been able to find a suitable pre-existing tool though. Bookwyrm looked interesting with its ActivityPub social features, but AFAICT it's limited to books only. I also looked into Koillection which was much more flexible, but doesn't seem to have any import ability for my pre-existing data.
Does anyone know of anything that might suit my purposes? I would like if it could automatically download cover art and I am ambivalent towards social features. I could bash together something simple myself but I'd like to avoid that if possible.
- Zabbix: for monitoring my servers
- Uptime Kuma: for monitoring whether sites are up
- Mattermost: for some communication, also automation (e.g. notifications from Uptime Kuma)
- Portainer: for managing Docker Swarm or Kubernetes more easily
- Gitea: for my source code
- Drone CI: for my CI/CD needs
- Sonatype Nexus: for storing everything from my code libraries, to container images
- SonarQube: for static code analysis in my projects
- OpenProject: for managing my projects and some documents/Wiki
- BookStack (now retired): for a knowledgebase/proper Wiki
- NextCloud: for my file management needs
- PeerTube: for my video hosting needs (currently mostly backups of stream VODs, because of bandwidth limitations otherwise)
- Grav: for my blog, over at https://blog.kronis.dev
- Yourls: for link shortening, because the blog sometimes breaks with query parameters otherwise
- Matomo: for analytics on the sites that I develop (cookieless currently)
- BackupPC: for my backups (incremental and deduplicated)
- my own mail server, for automation and as a backup private account
- some other sites that I've developed with the above tools: like my homepage, a forum or two and so on
- oh, and some game servers (like Minecraft) for friends
Historically, I hosted a few more pieces of software (like GitLab and GitLab CI), but I've both cut down on some and added some others over time.
Currently some of those run on my homelab (just regular Debian/Ubuntu boxes), whereas some are in the cloud, from a relatively local cloud provider. The total costs of the VPSes are around 425 euros a year, or around 35 euros a month. Currently the cluster has just short of 64 GB of RAM and 12 CPU cores total, over half of which are in my homelab.
There is some maintenance involved, but thanks to Docker containers and stuff like Let's Encrypt it's all tolerable and cheaper than cloud offerings (mostly since I don't care too much if anything goes down for a bit).
Edit: oh, here's something fun. I use Apache2 as my web server(s), instead of Nginx/Caddy/Traefik, because it has bunches of modules and mod_md is pretty okay for managing certificate renewal https://httpd.apache.org/docs/2.4/mod/mod_md.html
Apart from that, the combination of Portainer + Docker Swarm is a really nice and lightweight way to run containers, I'd say even a bit nicer than Hashicorp Nomad: https://www.portainer.io/ (though lightweight Kubernetes distros like K3s are okay for personal usage)
Also, NextCloud is really nice (for my needs), though some prefer Seafile to it: regardless of what you use, freeing yourself from a cloud storage subscription is nice, provided that you think of backups and such.
> I'm also curious how you decide to self host vs host on cloud. Is it purely for public facing vs private stuff or is it different?
For the most part, it's about the uptime.
I let my homelab servers sleep roughly when I do. They only draw 100 W from the wall due to 35 W TDP CPUs, but electricity bills are still a thing. Plus they do generate some noise, even with passive cooling for the CPUs.
But they're great as CI nodes or backup nodes as well, especially due to the fact that a 1 or 2 TB HDD is pretty cheap compared to most cloud offerings, even if you get multiple drives for backups and some spare ones.
Also, it's nice to host the public stuff in the cloud not just because of uptime concerns, but also because of a residential connection possibly being overwhelmed under load.
> Also, what are the majority of your costs? Do you have a few different VPSs or one big shared one?
I think either approach is valid.
Way back I had just 1-2 VPSes, at one point I had around 10 of them, then I gradually moved some of those into VMs in my homelab before consolidating them down further. Those VMs lead to a lot of unnecessary I/O due to multiple OS installs, as opposed to just containers running on the box instead.
As for how exactly I split things across nodes, I guess I look for vague categories like: this node will primarily host websites, this one will have most of the apps for writing code, this one will be a development node and will focus on CI, this one will mostly concern itself with backups.
Now the total amount of nodes that I have is under 10 and will probably probably remain that way. Thankfully, the cloud provider that I use does static (predictable) billing for a set amount of resources, so no AWS/GCP/Azure horror stories, as long as I only need VPSes instead of managed services.
I've also used Hetzner for when I needed more powerful nodes for a shorter amount of time (they don't do up front monthly billing): https://www.hetzner.com/
There are also some better known options that were a bit more expensive when I last used them: DigitalOcean, Linode, Scaleway, Vultr and so on. I guess it's about finding a good fit of features vs pricing and other factors.
For when I want to keep track of stuff that's connected directly to a particular codebase that I'm working on, turns out that Markdown files within the project's repo are more than sufficient.
Nothing wrong with BookStack, of course, it just felt like I didn't need it as a separate system.
Mh, my actual desktop, witch is not an web-app but it's definitively self-hosted anyway. I have a home server who host various things, the only web app so far happen to be home assistant for my p.v. and small automation around it. The rest revolve around ssh, imaps etc, like my mail setup where the homeserver downloads my mails from a handful of remote imaps, index locally with notmuch and I sync against it with muchsync over ssh on two desktops and a laptop.
Another is an experimental an obscene (so far) video-surveillance setup a day I'll humanize a bit, that allow just video streaming, no web app so far.
The homeserver manage some of my backups (zfs + ZnapZend) and offer some files over webdav, no WebUI again for them...
Long story short: I feel no need nor no reasons to COPY the Big of IT modern web model where I can get FAR BETTER without a modern WebVM improperly named browser for legacy reasons. The personal/home/user power lie in classic Desktop computing not in trying to recreate a buggy and limited GAFAM-alike infra at home... If/when a day enough people will realize that....
I am only self-hosting my VPN. I prefer managed services more because I want my things to work. Like I had given tought to self host Bitwarden but the thing is Password Manager is a backbone and I can aford to loose it.
I am DevOps since 2+ yrs and I have been working in IT for 4+ yrs. I have seen multiple disastors so I am a huge advocate of managed services.
I just run pi-hole and Tailscale for now. Tailscale modifies the DNS config on every device, and lets you specify the pi-hole server as the primary DNS server on the tailscale.com admin console. Much more convenient than figuring out DNS config for the 4 different OS's I use.
I switched to BitWarden a few weeks ago from LastPass. I want to self host VaultWarden which is an API compatible rewrite. I chose VaultWarden due to smaller memory footprint (its written in Rust) and it includes some paid BitWarden features for free.
Using Tailscale, it'll be awesome to test my site on my phone using https://macbook:8080. It was also fun to run `ping iphone` on my server and it works, but only when the phone isn't locked.
You can also configure PiHole as a DHCP server that sets itself as the DNS server. This way you only have to configure it once and devices should accept it as the default DNS. Had to do that to sidestep my locked down ISP router config…
Ahhh, from the alternate universe where I need to get a stable IP for my server without using Tailscale.
So you just run a 2nd DHCP server, all clients broadcast to every device to find an IP address. In order for the client to choose your server instead of the router, the server has to be faster to respond with an IP address, correct?
At first thought this seems fragile, but then again dhcpcd is the longest segment of my startup time...
Not exactly. I disabled my router‘s built in DHCP server so that they don’t conflict with each other. Luckily it didn’t have _that_ option locked down :)
- build services (on-demand containers through Proxmox)
- minetest (Minecraft clone) server (for experiments)
- administration/bookkeeping (Elicon online, an old Swedish web-based system which I kept running)
- a virtual router to tie all this together (OpenWRT running in a container in Proxmox)
- Database services needed for the above (PostgreSQL, Mysql, Redis all running in a single container)
- Backup service (rsnapshot) with backup being mirrored off-site (and out of country) to my brother's network
- all the scaffolding needed for the above
...and more. All of this on a single DL380G7 "under the stairs", connected through gigabit fibre. Proxmox on the host, Debian on the containers, OpenWRT in a container as well. Some VMs for Windows-on-demand which hardly see any use. The machine serves my extended family as well as a number of public resources. Waste heat is used to dry produce. The machine hardly breaks a sweat. Power comes from solar panels on the barn roof.
* OpenMediaVault - Have this running on a 4GB RAM Raspberry Pi connected via ethernet to my router. Only accessible in-network. This is my solution to accessing my music from multiple devices in the house, and it works great! I use foobar2000 on my Windows machine and RhythmBox on a Linux laptop and the experience is seamless.
* Pihole - got a PiZero running this. Low-effort setup, and great benefits.
* Jellyfin (kinda) - I say kinda because I only start up the server on my desktop when I want to watch something on the Roku in the living room.
* Whoogle Search - Very fast privacy-centric Google front-end.
Also have a bunch of tiny self-made webapps and a PG database running on an 8GB RAM RPI 4.
- a FreeNAS with a bunch of Samba file shares and a Plex. I tried Jellyfin, because I got annoyed with Plex trying to force me to create an account on their cloud stuff, when I just want to use it locally. But the Playstation wouldn't play videos from Jellyfin, so I stuck with Plex.
On a dedicated server with public IP addresses:
- mail (opensmtpd + rspamd + dovecot)
- blog (made with Hugo, a static site generator)
- git (gogs)
- Nextcloud
- XMPP (ejabberd)
- VPN (tinc)
Each of those services is in a separate jail and the jail with the blog has an nginx that serves as a reverse proxy for all http-speaking services.
I'm considering replacing XMPP with Matrix (looking at conduit) and tinc with Wireguard. With the latter I might wait until FreeBSD 14 with in-kernel Wireguard is out.
I host the following services via docker on my pi4:
- Main home PPPoE router
- 5ghz AP
- A pihole + coredns setup providing, LAN DNS adblocking via pihole pointing to a DoT upstream, a remotely accessible DoT server pointing to the pihole, a reverse DNS server for my IPv6 subnet and soon a DNS server for my domains :)
- A bunch of PHP telegram bots and static websites
- gitea
- woodpecker CI connected to the gitea instance
- mirror-to-gitea to mirror the projects I've still got on github
- invidious
- jellyfin
- plex (don't really use it, I prefer jellyfin but most of my friends use it so I set it up to aggregate our libraries)
- librespeed for LAN speed tests
- navidrome (really neat music player)
- nextcloud (mainly use it for the mail app via zoho mail)
- tvheadend
- a wireguard server
- vaultwarden
- a basic transmission instance, still gotta setup a nicer *arr combo :)
- Email
- Web (external and internal) and blog
- Matrix for chat (federated with friends)
- Media server
- File server
- Backups
- Password management
- Property surveillance
I run everything on a single computer powered by docker. Every self-hosted project has a dedicated docker-compose.yaml startup file. The services that run:
I've been putting more and more onto my Synology by using the existing packages that are provided. I used to lean very heavily on SaaS services but the pendulum swung and the cost and bloat have pushed me to just have some basic capabilities that don't change.
Audio Station - to listen to my mp3 collection
Synology Photos - replaced Google Photos
Plex - to access home photo/video library on our TV
Synapse (Matrix), Mastodon (Fediverse), WireGuard, and static websites.
Syncthing kinda counts, being primarily P2P, I suppose!
I'm generally focused on self-hosting communicating things, slowly branching out into hosting my own media services. Right now it's mostly just Syncthing'ing around my FLACs, but I'd like to set up streaming for my low-storage-space devices.
At home: none, quite honestly. I do enjoy playing around with Linux administration occasionally, but I have limited free time and energy, and I don’t want to accidentally get pwned. I do have a NAS for backups, but that’s it.
At work: WordPress and an in-house ERP system. We also have a NAS for file sharing, but management wants to ditch it and go the Microsoft 365 route.
In the past, i have self-hosted numerous CMS platforms, Matrix homeserver, mail server, Samba server, etc.; probably not so unique among self-hosters. Lately, I have been thinking of revisiting whether to stand up my matrix homeserver again (just for my family and I), or if I just pay a monthly fee to a matrix provider.
I've been hosting a fork of Tiny Tiny RSS[0] since 2005. Moving it between servers and Postgres instances has been easy and I like having the old data in perpetuity.
I don't think there is any website I care about checking regularly that does not have an rss/atom feed. All wordpress (and pretty much every other blog engine) sites do, subreddits do, https://hnrss.github.io/ for hacker news...
That said, I personally only use rss for slow-updating sites (pretty much just for personal weblogs and the news/announcements for some projects).
Quite common, most things that I try to subscribe to have a feed, including mainstream stuff like NYT, Axios, FiveThirtyEight, and smaller independent blogs. When one doesn't exist you can often use a Twitter feed instead (for the moment), but I've only had to resort to that once or twice.
Pretty common if not necessarily advertised. I'll echo what others have said-- any site I particularly care about has an RSS feed.
I've written scrapers for some things that don't expose RSS to dump into the tt-rss database, too. Having the data sitting in a relational database makes it easy to interface with. (I wrote my own "podcatcher" since 2005-era tt-rss didn't have any such functionality and just use the tt-rss database as its back-end.)
It is probably less than years ago, but its still a thing. In fact i heard youtube and other big sites (e.g. Reddit, etc.) still have rss feeds. Someone also told me that there is a bit of an uptick....buit this kind of thing is likely difficult to concretely quantify. As a single data point, I'm a fan of sites with RSS feeds...so i will frequent these sites more, and would be williung to pay for sites to have such a feature!
Yeah, when i am deciding if i wish to return to a website to read more of their stuff, the very first thing that i do is search for an rss/atom feed. If they lack such a feed, then chances are quite high that i won't return.
I'm looking for a good source code search solution that has navigation (browser based), and lets me add local repos. I have used Sourcegraph before and would use that, but perhaps there are better alternatives?
TinyTinyRSS, Matomo (Google Analytics alternative) and Joplin (note-taking software). Also, since someone mentioned bots, I'm running Redbot for Discord, and PhantomBot for Twitch/Discord.
NextCloud, Home Assistant, Minecraft, Caddy for some Hugo sites/blogs, Unify Controller, Vaultwarden, traefik, Sabnzbd. Used to do WireGuard but now I use Tailscale.
Forgot to add AdGuard home and FoundryVTT, for a friend.
Btw, Caddy is very cool, just 2 lines of config per website with https. Oh and the server is Ubuntu desktop, so I use it as a VNC desktop also (via ssh, hidden in the Tailnet), easy to leave stuff open, use for personal things at work etc.
I was using Astral for a while to provide maps and combat tracking for the games I run, and it felt really easy to throw together encounters on the fly. When they shuttered I got a free year of Roll20 and I have not been liking the change - everything, to me, feels overly complicated and being able to quickly integrate things that come out of rollplay into the encounter feel like they take many more steps.
I've been testing Planar Ally (https://www.planarally.io/) and it is interesting but rough around the edges.
It's awesome. 5€ per year, but more importantly: overrides. When Alexa understands "famous artist", you can set it to instead always play artist X that you actually requested ;)
I self‐host mostly because local copies of things give me some privacy (sites won’t know what my IP is searching for), and it also lets me work easily when Comcast is down… which is annoyingly frequent in my neighborhood.
All of these machines are running OpenBSD, except the gaming machines and the HTPC.
• Outgoing Email: OpenSMTPD, with mandatory TLS. Since I’m the only one sending email from my domain, the outgoing relay is hidden behind my LAN and my DKIM keys never leave my network. Outgoing mail gets routed via Wireguard through a VPS so it doesn’t look like it’s coming from a residential IP block.
• Incoming Email: OpenSMTPD on my MXes, with MTA‐STS and DNSSEC/DANE so as many senders use TLS as possible. Delivers to Maildir on my LAN, which I access directly using mblaze over SSH (https://github.com/leahneukirchen/mblaze) and IMAP via Dovecot (which supports Maildir backend).
• Roundcube webmail.
• DNS zones: NSD running on two VPSes, slaves pulling their config via WireGuard from the master which runs in a VM on my LAN.
• Public webserver, with personal (public) homepage, Git repositories (clonable and browsable via CGit), photo gallery, files/images/random files when I need to share them by sending a link in IRC, etc.
• Matrix: Synapse for the server, Element for the client. Besides hanging out in Matrix rooms I use this for one‐on‐one audio calls with my friends (generate a link, send it to them, and chat through the browser).
• Pleroma, so I can interact with the Mastodon network.
• Apertium for text translation. The range of languages is a bit limited but for supported pairs it’s nice to avoid Google Translate.
• A home theater PC in my living room running Kodi, which pulls all my Blu‐Rays from a home NAS.
• A powerful gaming machine that uses Steam to stream games to either the HTPC or my Steam Deck. I only use this at home… I wonder how bad the latency would be if I connected to it when on a trip?
• My music collection, whether ripped from CD or bought digitally, is automatically tagged and sorted with Beets, and I run the web plugin to access it over the web. Beets’s web interface is kind of primitive; I would love to replace it with something like FunkWhale.
• Full mirrors of websites with free content: Wikipedia, Wikimedia Commons, Wiktionary, Stack Overflow, Project Gutenberg, Standard Ebooks
• Full OpenBSD package mirrors
• OpenStreetMap, running OSRM (routing) on top of an open source Leaflet/Mapbox demo I set up years ago. I’ve been meaning to update this to something more modern and less reliant on Mapbox software.
• Radicale for CalDAV/CardDAV, so my calendar and contacts are synced across all my devices automatically.
• Home adblocking with Unbound (what most people use PiHole for I guess). DNS lookups for my home network are anonymized with DoH over Tor (CloudFlare provides documentation for how to do this).
• Ways to access my home network when away from home: WireGuard VPN in a roadwarrior configuration; public‐facing SSH (with WebAuthn‐backed keys); failing that, an HTTPS proxy with Squid. (Yes, I have been stuck at conferences where the wifi network blocked SSH, WireGuard, and all traffic that wasn’t HTTP/HTTPS or DNS from the blessed server!)
I know, it is a very low user count, the interesting bit is, I wrote this bot in Rust using async with Tokio, and it's running on a Raspberry Pi 2011.12 with 256 megs of ram and a single ARMv6 core.
I was able to compile a fully modern language for a 10 years old architecture, and on avg has a load of 0.2 of CPU usage with 10 megs of ram!
It was a very painless process using cross-rs, and I'm still quite surprised how well has worked until this day.