Here’s a business idea for someone: find the best open source/self hosted business and start up tools (such as a chat app, a project management tool, a wiki, etc) and offer startups a managed hosting service of those selected apps. Toss up a cheap but decent vps per client, run some open source monitoring tools for yourself, price as a flat rate per app under management (vs scaling rate of a per user seat in most SaaS products).
Seems like a pretty damn easy lifestyle business that many early startups would find valuable because it saves them money and the headache of doing it themselves.
Check out Cloudron: https://www.cloudron.io/
I think flat rates either make you too expensive for a small/scrappy operation getting off the ground, or too cheap to be able to offer what a large client might want (without additional agreements). In practice, this feels like the space that individuals doing sysadmin/IT stuff for local businesses are covering to me, but with even more caveats.
I think odoo is doing something like this. They offer a lot of different apps, from e commerce to project planning, to marketing to HR. But it’s priced per user.
I love, love all the energy in the ‘self-hosting’ space. These apps are all that stand between us and the cloud hegemony goin' on.
I’m interested in the alternative YouTube front ends like Pinchflat and Glance mentioned here. Ideally I’d avoid scrolling the YouTube homepage while still giving views and likes/subscribed to the creators I watch, because it helps support them (to what degree exactly I’m not sure but it seems like the right thing to do). I do find the archiving/downloading case meaningful as well, since a few yt vids that I enjoyed have been taken down over the years.
But.. reducing mindless scrolling while still giving me a way to follow content I care about seems good.
Does anyone have experience with using these? Are there alternate tools which are better than the ones listed here?
> But.. reducing mindless scrolling while still giving me a way to follow content I care about seems good.
My solution [1] to this was to create a static site that is built by a Github Actions workflow that runs every hour. The script just pulls RSS feeds I have listed in a .js file, and uses that to build the site [2]. The result is I'm more deliberate in what videos I watch, and I discover new creators organically (a friend recommends one, or I find them while doing a search).
For "favoriting" videos, I just add them to a folder in Firefox bookmarks manager.
My newest "feature" is a "Picks from your subscriptions" thing that uses an external Deno service [3] to grab a random video from one of my random subscriptions. This helps me discover old videos from my subscriptions I may have never seen.
[1] https://github.com/kevinfiol/youtube
I use pinchflat mostly for archiving channels that occasionally take videos down. It technically has an inbuilt video player, but that's more for "debugging". The intended non-archival usecase is have the download folder picked up by your Plex or Jellyfin and watch it there.
Of course that deprives you from interaction with the actual youtube content. But for support, $1/month on the creator's patreon goes further, and for engagement you can join the creator's discord. The only thing you are really missing out on is feeding the youtube algorithm.
> (to what degree exactly I’m not sure but it seems like the right thing to do)
It doesn't help them very much, or at all. If you want to support them, you can pay them.
The idle sleep feature of Go Proxy could be a gamechanger since cheap servers can host lots of infrequently used apps. Guess it should become the standard and extend to background processes too, like a PM2 or Systemd.
https://github.com/yusing/go-proxy?tab=readme-ov-file#idlesl...
How is go-proxy implementing idle sleep? Swap and/or CGI can do similar things, but maybe go-proxy uses knowledge of QPS to know which services to make idle?
What’s the purpose of appending ?ref=selfh.st to every outgoing link? Generally you can’t just make up parameters for other people’s websites.
To allow the maintainers of the projects to see in the traffic analytics where an influx of stars comes from, I guess
https://docs.github.com/en/repositories/viewing-activity-and...
Surely that must use conventional HTTP referrer data—for which we have well established standards, decades of experience managing interactions and edge cases, and norms for respecting user consent—and not some apparently wishful query parameter?
https://httpwg.org/specs/rfc9110.html#field.referer
A user agent SHOULD NOT send a Referer header field if the referring resource was accessed with a secure protocol and the request target has an origin differing from that of the referring resource, unless the referring resource explicitly allows Referer to be sent. A user agent MUST NOT send a Referer header field in an unsecured HTTP request if the referring resource was accessed with a secure protocol.
Yes, that’s the part about interactions and consent. The intent then is to circumvent that? A case of those pesky matters of security and respect for human dignity getting in the way of SEO?
I’ve found Home Assistant and i’s add-ons systems a good way to self host without the fuss. If you need something that is not there you can create your own repo and it’s pretty easy to create your own add-on collection
I'm a very satisfied home assistant user, but I don't run much in the way of addons. Any you'd recommend?
Self-hosting is such a crapshoot. The installation/setup process is completely bespoke for each different app. Self-hosting apps could have a real impact on overpriced SaaS if installation could be made as simple as installing a mobile app.
I self host about 30 services. Each of them is in a docker container and I do not know what happens inside.
I only configure the startup in a YAML file and I am done. Works great.
This is a blessing and a curse. If there is a bug or something that you need to debug in it, it can be painful
I don't find debugging in docker containers difficult, but maybe that's only because I've been doing selfhosted stuff in docker for a few years. Usually exec'ing into a bash she'll inside the container satisfies my need to easily run commands inside the container.
What about debugging is difficult for you when docker containers are involved?
Most of them are OSS so you could make a real impact if you spearheaded this! Some buy-in on some sort of standardized install process from the top projects could go a long way if you decide to pursue.
I’ve had enough success with using docker images (lots of projects provide them) that I don’t feel the pain acutely, but there certainly is some bespoke fiddling with every one. To me that’s part of the self-hosting experience, but I know plenty of people that don’t self host because of it, can’t blame them.
There are many tools that provide this experience but you then lose some of customization.
Some examples are
Cosmos (https://github.com/azukaar/Cosmos-Server)
Younohost (https://yunohost.org/)
Runtipi (https://github.com/runtipi/runtipi)
Also most of the popular NAS setups. I'm pretty happy self-hosting everything on my Unraid server. I just checked, all of the tools in the article are in Unraid's App center (except Streamyfin which is a phone app). And if something isn't you can add docker containers or VMs as necessary.
Thanks for the list. I hadn't heard of some of those.
I feel so old, I haven't heard of any of them.
When I think of installing software, my mind goes to things like "apt-get install" and maybe opening /etc/xyz.conf in "vi". Who knew installing software needed so much support? Even docker seems so overboard for 90% of "single home lab machine" use cases.
That apt-get install never actuay worked properly - what if packageA uses version A of a dependency, but packageB uses version B of that same dependency? Sure, debian/Ubuntu maintainers did the painstaking job of making an environment where most packages can happily live together, but that only worked until you had to use this new version of program C that requires a newer dependency for this and that and that.
Also, it always left behind orphaned packages, a good test was/is to install the whole of gnome, then the whole of plasma, and remove them both. Ideally, nothing should be left alone.
The only package manager capable of that is Nix (and those that use the same fundamental model). A second best option is to simply ship your whole work machine, a la docker.
Pretty much everything has an official docker image and included docker-compose file at this point, so installation is pretty standard and can be done in a couple of minutes for just about any of them.
But if you do want mobile phone app level of simplicity, there are options like Cosmos that handle everything for you and just give a pretty UI.
> The installation/setup process is completely bespoke for each different app
For the majority of them, there's a docker container, often with example configs. It's not that far removed from installing a mobile app and then having a wizard ask you how you want stuff configured.
As far as I know, there are no inherent problems preventing simple installations. You make it sound like there is.
No inherent problems, just a very fractured landscape where install could involve: Linux, MacOS, Windows, WSL, Docker, or AWS. If your setup has anything that differs from the developer, it could unexpectedly take hours of extra troubleshooting.
Most self-hosted services I've seen all use Docker, which have nothing to do with your OS or cloud provider.
The biggest hurdle for me is being too lazy to setup proper DR/backups.