ProfessionalAd3026

joined 11 months ago
sorted by: new top controversial old
What service one should NOT self host in c/[email protected]
[–] [email protected] 1 points 10 months ago (3 children)

A CDN

NGINX RTMP Streaming Server for Church in c/[email protected]
[–] [email protected] 1 points 10 months ago

I had a similar setup initially due to the same reasons. I only used HLS-based playback. I would chop the files locally and upload them to our web server. I learned the hard way that Android mobile devices are a nightmare to support. Some had simply no sound. Significantly, older people would not care if a device is still supported. Some simply wouldn’t load the sound. Eventually, we moved to YouTube.

The solution we kept, though, is something you might want to consider. We have our lifestream on an internal web page. If it’s close to the actual service, we have a countdown and lifestream. Once the lifestream is done, the same page switches to VOD. If you didn't make it to the real lifestream, why do you need to be on time for the re-stream?

Security question about hosting custom service in c/[email protected]
[–] [email protected] 1 points 11 months ago

Look into certificate based authentication. If the clients don’t present a certificate issued by your CA, the connection won’t be fully established.

Forwarding incoming email with an smtp as a service relay in c/[email protected]
[–] [email protected] 1 points 11 months ago

T-Online sucks for many years already. They wouldn’t even follow standard back in the days.

3
ZFS voes (alien.top)
 

So, I bought a nice Zotac ZBOX with 2TB local storage and 16 GB memory. Installed Debian with k3s and partitioned the disc to have a ZFS pool for the containers. I've set up the ZFS provisioner.

I'm migrating away from an RPi4 with 4GB of memory and docker-compose

I run out of memory and have my pods evicted while not having all my things migrated. Okay, I've got two new things: ArgoCD and Navidrome. But what the heck?

Now I've updated k3s according to the manual (just download the binary and restart). It wouldn't start. Ran the installer instead. During boot-up, two of my ZFS volumes got deleted by the provisioner. F*ck. Luckily no important data on there, just extra work to redo the migration. That's the second time the provisioner kissed a volume goodbye. The first time, I was to blame because I messed up a commit in my ArgoCD repo. Data was faster gone than I could say bye-bye.

I work with Kubernetes daily. K3s seemed a logical choice for managing a more significant environment. I looked into Proxmox, but docker containers are not supported.

I spent every free minute for the past 14 days migrating, but I have my doubts about the correct path. Eventually, I need a stable system.

ZFS was supposed to save me from some of those stupid mistakes. I run Btrfs so far, and snapshots are a lifesaver. Can't wrap my head around them in ZFS. Unfortunately, is the Btrfs provisioner not very widespread/stable.

Any advice or encouragement?

PS: Velero is still on the todo list... :(