That’s odd because the clients are just web apps I think. That should work without crashing on a stable OS. I use them on Android mobile and Android TV with extensive subtitle usage and haven’t seen instability.
A funny thing I noticed is that the client distributed in F-Droid is extremely old even though it says it’s updated recently.
I switched from a heavily used Plex server with about 10 users to Jellyfin with the samw usage patterns abour half a year ago. So far it’s been pretty smooth sailing. A better world is possible!
Hm, I thought Chromecast needs HTTPS and internet-visible endpoint.
Oh? I haven’t upgraded yet. 😄
That’s how it works with Tailscale as well. Tailscale creates Wireguard tunnels underneath between the different devices. There’s also an open-source self-hostable Tailscale control plane.
I do use both a reverse proxy and Tailscale. All services are proxied. All services except for Jellyfin are accessed only via Tailscale. Jellyfin is publicly available. I’ve obscured it a bit by setting up long, randomly generated DNS name. The proxy would only forward traffic to Jellyfin if the request comes from that exact DNS name. Bots would have to know this name for the proxy to entertain their attempts at all. Then every user has long, randomly-generated password. I prefer to only use it behind Tailscale but some of my family needs direct access. Also Chromecast.
I do. No issues.
Switched to Jellyfin after more than a decade with Plex. Prettey… prettey… pretty good.
Can confirm it’s the truth.
It’ll probably be great. That’s pretty much giaranteed based on the physics. That said, if you already have a mesh that has good connectivity, an antenna like this would probably just reduce the latency a bit. If you have latency-sensitive applications. I use an HA Yellow with its built-in Zigbee radio. It only reaches 5-10m. Everything after that is connected through the mesh.
Never had it. Z-Wave needs a differrnt antenna and since they’re using full-size antennas for these devices, the Z-Wave one is significantly larger so it’s a separate device.
Got the module some years ago when there were massive shortages. I found a couple CM4s and bought them at the time.
What are you comparing it to?
You’re supposed to set SQM lower than the WAN throughput. I think you’re right that by default it limits it by about 10%. There was some study over a decade ago on this that showed 20-25% limit is best for maximizing responsiveness under load. It’s not possible to effectively schedule packets if there’s no headroom.
I have a few Pi 4 + UE300 routers in operation that work just as well but this is a nice alternative if you have a CM4 lying around.
I don’t think so, not by DFRobot at least. That said, I think 2.5G is only useful for >1G internet connections. On the LAN side, the switch is what matters for LAN throughput.
Nice to see you here Mr. President.
ZFS. It runs on whatever RAM you give it.
As someone who’s done cloud infrastructure professionally, this is the right way to make a project for setting up self-hosted applications. Not writing a bunch of bash scripts and putting them behind some web UI. We have well established infrastructure/config-as-code systems that are the gold standard which runs most clouds out there. Ansible is one of them. That’s the right tool for this job and a ton of professionals understand it and therefore can easily contribute improvements for the ones who don’t to use. I’m unfortunatrly invested in SaltStack but I wouldn’t feel worried to deploy a (well reviewed) project built on Ansible. Then slap a web UI on it if you like but that should be another project that hooks uses this one.
This is cool but for self-hosting you probably want a more robust monitoring system capable of alerting at all times. Prometheus is what I use. It also gathers data over time and can monitor many machines.