It’s all in the title :) Personnally, i’d say my old Cisco Catalyst 2960. Totally EOL but it’s my first managed switch.
the fact it holds together and is “reliable”.
That’s me! The fact that I don’t have to deal with it on a daily basis 🥰
Shits reliable. I have automated 3-2-1 backups. PSU supports power to router and lab, and due to NUC tech, it stays on for 5+ hours on battery. Anything that could be automated is automated.
it stays on for 5+ hours on battery
Dang, that’s impressive! What’s your typical power draw and would you recommend any UPS units to look into (or stay away from)?
It’s around 15W! I use a rack mounted APC1500. German company, seemingly very reputable. I bought a used unit for about 100 bucks. It’s way too overpowered for my server, so I hooked my workstation, router and switch to it as well. It gives me enough time to theoretically save up my work or close any games (which is also half automated, auto shutdown in 30 seconds unless pacman has a lock up) and shut down and still have 60-70% battery left.
The fact that it is a 3-node proxmox cluster with high availability. When one node fails, the other two pick up all the services the failed node took with it.
And it is all in a 10inch rack.
So w/ a three-node cluster, while all three are healthy does it use all three nodes like a load balancer type deal, or is there a primary node that does all the work while the other two nodes are basically chillin’ until the primary node fails?
Usually (as in, when not configured otherwise), all 3 nodes are equal and the containers and vms are load balanced across all of them.
You can however set priorities for each container or vm, you can also set affinity or avoidance rules if some containers should run on the same node or not.
So both is possible.
I have them set up to load balance dynamically.
deleted by creator
My main server is a puny Raspberry Pi 5 with no disks that still packs a lot of services for my home.
My other server is an ancient 4 core box with an equally old graphics card and a bunch of disks (some spinning, some solid state) arranged in a ZFS array. It handles backups and running a local nasty tempered LLM modelled after GlaDOS.
I’m proud it’s run in three countries and been accessed from four, depending on your definition of country
What do you mean by country? I would have thought everyone had the same definition of country
I mean the China/Taiwan issue. Taiwan is definitely a country but they are/were also the fascist losers of the Chinese Civil War. China says Taiwan is a part of China, Taiwan says they rule all of China. That’s what I mean.
deleted by creator
I have no idea what this means my South American bro
deleted by creator
I’m always doing that too, lol!
Because only South Americans fight the invisible enemy that is everyone in the world calling people from America, Americans. Making up words like USians makes it an easy guess, haha
I’m in the same boat and also curious. Ping me when they answer please 🙏
64 days of continuous uptime today!! I know this isn’t much compared to serious hosting, but I started self-hosting not so long ago and it’s my record so far.
Apart from that, hosting my own messenger (Matrix) instance and actually having my friends switch to it from proprietary platforms. The sense of truely owning my data is deeply satisfying.
The 5 HDDs that are playing so nicely together with ZFS, rock stable now for 8 years or so.
I have 3 machines running proxmox, and if one of them dies, the other 2 can pick up the slack and spin up the LXCs with the docker services to prevent important services from going down.
Automatically, within a couple of minutes, with little to no data loss.
It was a pain, it maybe is not super reliable, and it’s all running on older hardware. But it is running, with good uptime.
My old laptop (also EOL) still running as my server I guess ? I’m still looking for some time to switch to my old desktop with 5TB.
Not much but it’s kinda needed for my curreent Arr stack and my self-hosted services i want to share with family members (SFTPgo, vaultwarden, invidious, baikal, navidrome, jellyfin, seerr…)
Still need some time to make our network more reliable :) I still have 2 managed swiches to put into our network !
Mann so many things to do, but not enough time in a everyday life :( !
Still need some time to make our network more reliable :) I still have 2 managed swiches to put into our network !
Would you care to expand on the network reliability part? I’m curious what kinds of problems you’re looking to address w/ the managed switches that you aren’t using yet. For example, will they help in the event of a power/ISP outage?
Currently? 4-5 months “playing” with local models and in the last week or two it has REALLY started to come together, and I have NEVER in like 25+ years of having a home lab/server(s) under the desk, but as happy or impressed. Fully offline and private, scratch-coded agent with a boatload of models behind it, spread across five GPUs in four VMs, and now with fully private/local (also scratch coded) voice interface. Absolute most incredible thing I have put together in my life, and that’s not a short list of ‘cool shit’ I’ve built over the years.
Historically? Having such high-end networking gear to ‘play’ with in the early 2000s that I actually sold and shipped some to a small ISP overseas to use for production 🤣
