Oh I meant the running in a vm part. I sell hardware to replace the pi if you wanted too, but you may already have something else suitable. Virtualizing/containerizing makes a lot of sense for isolating services from each other, at the expense of complexity to allow for higher reliability and better tooling (backups/version control).
If umbrel can’t do it for ya, you can learn to do it for yourself. Check out proxmox & install it on a spare machine.
I've been droolong over the Start9 [Server Pure](https://store.start9.com/products/server-pure) - any thoughts on that?
All around a solid device/price/company, I just would still rather see proxmox running in between bare metal & application - you get extra tooling that is useful.
I was literally just looking at proxmox site. Ty
Not until I’m running my own lightning node & can communicate to it without DNS. It’s a rough project 😅
wireguard?
i use it to run a reverse proxy on a VPS that is pointed to by DNS, but actually accessing it is via the IP address over wireguard
iit has other benefits too... it proxies my mobile phone and my laptop/bitcoin node rig to a location in romania which means i'm not subject to local network restrictions (vodafone, at least, blocks many sites in obedience to the pt mafiya
if you aren't already using it, it's a great way to make remote access possible for any other device connected to the same wireguard server
i dunno about failovers in case the wg proxy server goes down or offline though... but possibly it can be set up so you have multiple wg proxies that automatically try a backup if the main one is failing...
seems to me like this is something that wireguard really needs in its configuration - probably is just some detail for routing metric that can actually be put in the config, just not sure if a less priority one takes over if a greater priority one fails
i'd quite like to have a second VPS to back up my first, that also runs a regular sync to the primary, and put a DNS round robin record as well... but hah... yeah... probably more work than it's worth if the outage is brief, which mostly they are, at worst half a day
well maybe you can find a way to mod/build that profitably.. it is very useful, but i think it does raise the cost of the hardware/reduce performance a little handling the hypervision
also for that it might be better to use LXC containers anyway (via nspawn) that is extremely light weight and simple and gives you a lot of the same capabilities (idk about proxmox but libvirt can run these on the backend too)
Yup, you can access lxc’s in prox. Better than a VM for this for sure
Or start with incus LXD’s fork and not run by Canonical
not sure - seems to have a bit of extra overhead for nostr:npub1r7psmkr4zv93xnal8un6d8hvmpsn5jvhfzn3kk38rfcel6awznks7znspg 's use case as well as a new API to learn (vs proxmox)
Have you seen the site in my bio? 🤓
Start9 & I sell practically the same used hw, minus a slightly better cpu in their favor.
The performance hit of a vm on these is negligible, especially for users coming from a pi. Virtualization costs less than 5% on modern hardware, containerization even less.
