-50
The 'if this goes down, I riot' self-hosted app (sh.itjust.works)
submitted 1 year ago by to c/selfhosted@lemmy.world
8 comments fedilink hide all child comments
 

alright folks, let's get real. we all have our sprawling digital fortresses, carefully constructed brick by brick. but there's always that one piece of software, that one perfectly tuned instance, where if it so much as hiccuped, you'd be ready to throw the entire homelab out the window and start fresh in a cave. what is it for you? what's your absolute, non-negotiable, 'i will personally visit the datacenter if this fails' self-hosted application? for me, it's my media server stack. my wife would disown me. don't let me down, arr suite.

you are viewing a single comment's thread
view the rest of the comments
[–] 4 points 1 year ago (4 children)

Well the only way my media server goes down is if the power is out for more than a few hours or my internet connection is down.

So I'd have to say that.

  • permalink
  • fedilink
  • source
  • hideshow 4 child comments
    •
  • [–] -1 points 1 year ago (3 children)

    I don't believe you, but I'd like to be proven wrong.

    I expect you have a UPS that feeds your hosts and networking equipment and something like ZFS for disk redundancy. This protects against the most common failures and is usually enough, but there are still single points of failure in such a setup, that are not as common, not as hard to deal with through manual intervention, and quite difficult to protect with redundancy.

    I would be surprised if you are protected against the following single points of failure without manual intervention:

    • NAS machine (not just disk) failure. You would need to have a multi-node distributed storage, like Ceph, to protect against this.
    • Networking equipment failure. I think you can do some magic with BGP to do this, but I'm not a network engineer and I've never set up a redundant network.
  • permalink
  • fedilink
  • source
  • parent
  • hideshow 3 child comments
    •
  • [–] 3 points 1 year ago (1 child)

    Ceph for the proxmox cluster, 2x48 port switch + 16 port 10gbit as the core, 2xNAS (technically one is the backup, and there would be a few moments of downtime as the containers restart - a different container with the same config pointed to the backup NAS instead).

    UPS and internet are the SPoF.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 1 child comment
    •
  • [–] 2 points 1 year ago*

    One thing I'd add is a whole house surge suppressor.

    I saw the power lines arcing to either each other or the bamboo outside our house last week during a bad storm.

    A whole house surge suppressor is only like $100, I'm gonna get one soon and install it. I saw it's best to install it as close as possible to the main incoming power lugs, one lead on each leg of the split phase 120/240.

    A UPS will protect against surges but it's just a good idea with how many appliances and devices have circuit boards in homes these days. Like your furnace, oven, washing machine, game console, TV, etc.

    I had an insane surge last winter so it's a long time coming haha. I woke up and half my circuits were off. I measured 170v to gnd on one of the legs. Power company and fire dept had to show up to fix it.

    Power is ehh not great where I live.

    Edit: for your point about a NAS failure. If that were to happen, since I use unRAID, I could just throw the disks on any Linux PC and my data would be fine.

  • permalink
  • fedilink
  • source
  • parent