Let’s do another round of this. What’s going on on your servers, any cool things? How’d the recent CVE flood tangent you and did it change how you see/do stuff?
You must log in or register to comment.
Everything behind a VPN now, Wireguard is my single point of failure…
I’m considering trying to host my own mail (receive and internal send only, at least at first), as I have a friend who’s hosting all the useful services and mail’s the only one left.
Also experimenting with (rootless) podman, I’m finding the permission mapping a bit confusing tbh.
I have electric baseboard heaters so i like to heat my home with my server in the winter. I dont use it enough in the summer to warrant the power draw, but i’m looking forward to doing so this winter.
I’m hoping the AI bubble pops soon… i’d really like to pick up a decent accelerator card for AI for my desktop. My current server doesn’t have the PCIe slots left (all filled with U.2 drives on PCIe adapters or HBAs) so it would need to go in my desktop… i thought 64GB was a lot when i built this 3 years ago but now i’m wishing I put 128GB in it…
My server has 238GB though (ordered a 256GB kit and one stick was dead. Caught that on the 4th test cycle with memtest86+ pro which I 100% paid for.) Since I use it for not only ZFS but also all of my random containers and game server VMs.
Finally got around to self hosting Fluxer (Discord alternative) because they’ve implemented an easy to use setup.
Now I’m trying to optimise the setup for my needs as a backup space if my friends decide to stop using Discord and it’ll be a frictionless changeover.
On a side note; it’s the most containers for a single hosted service that I have. I can’t check the exact number right now but it’s over 10 containers!
Got a static IP routed in from https://getrealip.net/ so I can go back to hosting my own servers at home again properly without needing to hack up everything to route through Cloudflare with Tailscale. CGNAT is really messing up the internet. The service even supports reverse DNS, yay.
Ordered a cheap Coral TPU for my server (an older HD mini PC with AMD CPU, no discrete GPU), just waiting on it to show up.
Last week, I finally got my Tapo cameras + Frigate + Home Assistant all working together, but now my server’s CPU idles between 65% and 90%, hence the TPU.
Decided to get back into podcasts recently, but baked-in ads and sponsor reads make standard listening unbearable.
I spent the weekend spinning up MinusPod, an open-source tool that detects ad reads and strips them using ffmpeg. It uses Whisper for local transcription, then feeds the transcript to an LLM to identify exact timestamps for sponsor segments and commercial breaks before slicing them out. I routed the text processing through Gemini Flash, which runs about $0.07 per episode. Clean, ad-free audio feeds with zero manual editing.
This reminds me of a project idea I’ve had.
Auto-muting a TV using an HDMI splitter and an SBC running some sort of algorithm that detects if there’s an ad on the TV.
iSponsorBlockTV?
Isn’t that only for YouTube?
But yes, something similar to that.
A while ago I made my first attempt at setting up a SSO for some of the things I host. I figured out things that have built in support, like Immich, but struggled with getting it done through Proxy for things that don’t have baked in OAuth or OIDC support.
I revisited it yesterday and figured everything out so I’m happy about that. Navidrome was one of the more difficult ones to figure out because I need SSO authentication for the web-ui, but username/password for the API. That was actually pretty easy once I wrapped my head around it.
I picked Authentik based on general positive things I’ve heard about it, and I use Nginx Proxy Manager for reverse proxy.
I’m also on authentik, because of the extensive list of tutorials. I’m also using Navidrome but couldn’t be bothered to set up SSO there since I’m mostly listening via clients
I finally ran a hardware accelerated ML container for immich.
My immich is in a VPS and my GPU is in my home server. Cheaper to run that than to rent GPU compute.
Thanks for this post, something I miss from earlier selfhosting communities.
I have simply running two SFF Lenovo Thinkcentres—one for all local services and another for all online services. They could run on one but I keep them separate for security and stability reasons. I often crash the local one by running out of space…
I finally got to building a second backup recently which I am happy about.
Honestly, since moving and downsizing from several services at home and at work, this solution works. I run out of space frequently but it forces me to not hoard things I will not actually use. My previous rack server seeded something like 50tb as I kept everything I leached to benefit the community, now less possible.
This has also made me move away from services that are not needed.
-
[ ] I would like to offload all my home videos to an external drive as I am running out of space for actual services and Jellyfin media.
-
[ ] I would like to consolidate all of my local and online databases rather than run a new one for each container stack.
-
Feeling better every day having my stuff under my control and away from prying eyes of trackers and other unsavory stuff.
After the recent plex player debacle I’m once again trying to come up with a way for me and my diaspora to share media using Jellyfin, but without relying on Tailscale. We have several media libraries, and some of the library hosters are not too tech savvy, so coming up with a solution to replace the ease-of-use of plex hasn’t been easy.
Maybe try netbird?
What’s wrong with tailscale?
Hey, I did the same. It’s not as bad as it seems. Get yourself a domain name from cloudflare and use it. I got mine for $10 a year. Then just installed cloudflare on my Debian server and all that (I’m sure you know the rest) then made users on my admin jellyfin dashboard and started sending people their usernames and passwords. The only difference for them is now they would need to enter the server URL on the first page of whatever they use then username and password. Good thing these jellyfin clients have the page of entering the server URL in your face with their “connect to server” button. Let me know if you have any questions.
Thanks for the details. I don’t think that covers my main problem though: letting other people who are not tech savvy share their libraries with me.
A user story would be: “I want to enter a search and find the result across all my friends’ shared libraries.”
Careful with using CF Tunnels for streaming content. It’s against their ToS to use the free tier for streaming content, so they can technically turn it off whenever they want from my understanding.
Against whose TOS? Jellyfin’s? I’ve never heard of that. Do explain why please.
Edit: I re-read your comment a couple of times (English is my second language), then searched some stuff. I didn’t even know cloudflare did that. I’ve been using them for over 2 years with no issues. I guess I don’t stress their servers enough? lol. If they stop me, I’ll just move to another company. I won’t even bother asking them or anything.
Against cloudflares ToS
Thank you. I realized that after reading the comment a couple of times. Yeah, they can screw off. If they stop me, I’ll just move my stuff somewhere else.
From what I understand, if you turn off caching you aren’t violating their TOS. I’ve been doing this for about 2 years now and have yet to have anything happen. Multiple friends use my Jellyfin server daily.
My automatic weekly backup ran for the first time last night. Everything worked nicely.
I practiced restoring my offsite backup this week. Good experience.
Then I printed and laminated the instructions and stored in two locations.
Just curious. How long does lamination last? I assume light hurts paper.
I dunno. Longer than bare paper, I hope.
If it’s automated, why only run it weekly?
Some backup methods – like VM backups on a basic Proxmox VE – only support full backups, so they put a lot of stress on the system and take a lot of space. Even if the performance hit or service disruption during the backup is manageable, a homelabber has finite disk space for backups, so there’s a tradeoff between RPO (most recent backup) and retention (oldest backup). Keeping 2 weeks of backups is cheap for weekly, expensive daily.
Ideally, everything would be backed up incrementally at fairly short intervals (short RPO, long retention), and restorable very quickly (short RTO). Getting there is a journey and weekly automatic backups is a great step.
But maybe losing a week of data is fine in their case. There are parts of my lab that would even be fine with monthly backups…
(that’s correct, I just wanted to add that Proxmox Backup Server supports incremental backups)
Ah that makes sense, thanks!
I’m always super excited for this kind of stuff :D
The only PHP vuln I had was from parts-db which I deployed recently to check it out. Haven’t set it up yet seems like a lot of work. I keep a running record of my electronics parts in obsidian hoping maybe I can bulk add em or something.
Just completed my set of needed services by adding nextcloud to the list. Now life is good. Jellyfin: check. Navidrome: check. Pihole: check. Invidious: check. Nextcloud: check. What else? lol
You’ll find more stuff for problems you didn’t know you have. :D
