I’ve been using Authelia with several OIDC integrations for a while now. Works great. They’ve released a huge update like a day ago too. Out of the ones you listed, it’s very lightweight too. The docs are a bit all over the place but it is quite comprehensive.

I did look at Zitadel and tried setting it up myself but I just couldn’t get it to work. The docs are a bit vague.

If you just want to view logs, then a lightweight viewer I really recommend is Dozzle.

I’ve tried nearly every selfhosted dashboard out there and in the end settled for static html/css/js. If you want to access links quickly by typing abbreviations then use something like https://github.com/Ozencb/tilde-enhanced. A lot lighter and can be used with an existing webserver too.

https://tailscale.com/kb/1054/dns#nameservers

and

https://tailscale.com/kb/1114/pi-hole#step-3-set-your-raspberry-pi-as-your-dns-server

Set tailscale to use your dns server to resolve your services (or all traffic if you prefer). Assuming your dns server is on 100.x.x.1:53, then put 100.x.x.1 as a nameserver.

How about Uptime Kuma status pages? They’re separate from the admin page and you can add Docker containers as monitors.

SFTPGo supports OIDC and has a lot of ACL features. It allows users to have their own folders, as well as shared volumes between a group.

This might be an issue with opensearch.xml, which is a standard for how browsers recognise search engines.

See here:

https://github.com/hnhx/librex/blob/main/opensearch.xml.example

I don’t know how you’re hosting it, but when I was hosting LibreX, I had to make an opensearch.xml with the correct domain and bind mount it to the correct location. I don’t exactly remember the details since I moved to Searxng.

Also, if you’re not aware, LibreX was forked to LibreY, which is the updated repo.

Navidrome replaced Spotify for me, with Symfonium on Android, I’m never going back. On PC you can use any Subsonic client, and there are plenty I threw Tailscale on top to access it when I go out.

Org-mode, with Orgzly on Android, sync via a WebDav server, which you can also mount on you PC and literally use any editor to edit.

Okay I think I might know what you mean? I just tried doing that and got it to work. We can compare what we did. Here’s mine.

I created a shared folder called “Shared”

then I create a group called “All” and mount the “Shared” folder to /shared

I went to a user and add them to group “All”

Examining that user’s files

I can navigate into that shared folder and access everything (I have stuff in there already).

To set up the folder, which I called “shared”, I set the home directory for it to /srv/sftpgo/data/shared. For reference, my user home directory is /srv/sftpgo/data/user1. Then to allow user1 to access it, I mount it as a virtual folder. Is this what you did?

+1 for Netdata, very fast and a lot of alerts have already been set-up. It also has a lot of plugins, as well as the ability to use Prometheus metric endpoints. The local dashboard is near parity with the cloud one, and setting it up is as easy as running their bootstrap scripts. There is decent documentation too, if one gets stuck.

I think all the RAM related issues were closed a while back and were supposedly fixed. I just don’t understand why when interfacing with the front-end, it uses so much it would get OOM kill itself with 1.5 GB allocated memory.

Every page, as well as loading in the initial dashboard from an idle state, spikes the RAM. Are there no clever lazyloading happening or something? Surely viewing and modifying database entries can’t be this memory intensive?

Maybe it’s just an unoptimized Python thing. I stopped self-hosting stuff written in Python, with the exception of Linkding (which takes a while to also submit a link) and Whoogle.

https://awesome-selfhosted.net/tags/document-management.html#paperless-ngx

I stopped using Paperless-NGX for this reason. It eats RAM and CPU insanely even after configuring it to stop doing OCR and no ML. I wish there is a Go alternative.

I’m using Headscale with minimal issues. It’s low on resource and the docs Tailscale provides applies to it which is neat.

+1 for Dokuwiki. It’s simple and fast, has a flat file storage, and a nice syntax.

Wise move, all the default alerts that came preconfigured are such a timesaver. I realise what I needed was alerts and not really visualization.

Monitoring. Try out Prometheus/InfluxDB and Grafana, throw Loki in there too… It’ll keep you busy for a few days to a week at least.

I did all of that and I just use Netdata now.

I don’t know the arm architecture that well, but have you considered soulseek with slskd?

Edit: no metadata control unfortunately (I use Lidarr for that) but I find it nice to be able to get proper releases, unlike from YT and Spotify.

Not the answer you’re looking for, but I have a self-hosted Calibre server and I stuck to a second hand Kindle I got. It would be neat to be able to browse my remote library like on the Kobo, but I’d rather buy what’s second-hand, cheap and readily available (lots of these perfectly working pre-loved Kindles and Kobos). Transfer lots of books at once and I rarely have to do it since I read slowly. If you use it for magazines/news/comics, then other more libre and open recommendations seem quite good.