I’m currently traveling for months at a time and my homelab has become unreachable to me over VPN due to a unknown complication after a power outage.
Just as a learning experience for all, my mistake was that I set-up my VPN very far down the stack - as a wg-easy app inside TrueNAS SCALE’s apps ecosystem. My very important reason for doing it was that way was that wg-easy allows for setting up client devices with a QR code…
Anyway, the NAS is not booting back up nor do the TrueNAS apps. I should’ve set my VPN up right at the front of the network - on my MikroTik router that also supports Wireguard. The funny thing is I was so happy that my NAS has IPMI and whatnot but now I can’t even access it.
For now the NAS is kept powered on from what I know, it just doesn’t boot. This should help prevent bitrot until I’m back. All important files are backed up on a 3rd party service.
It’s a shame my Jellyfin and Navidrome inaccessible, but I’ll live.
Now I’m thinking about buying an UPS so that this doesn’t happen in the future. I’d like the UPS to be fanless and rackmount, so that limits me to ~700VA territory.
Devices in my homelab pull about 65W idle and spike to say 150W when everything is booting. ISP modem, router, POE+ switch, AP, NAS. I might add another 20W due to a Lenovo M920q in the future.
I only really care about NUT and graceful shutdown instead of long runtime on battery.
I was thinking about this: https://www.apc.com/us/en/product/SMT750RMI2U/
In my country I can get it with new batteries (no front panel) and a network card for NUT for a total of 180 EUR.
Would that work? Would you be afraid of leaving an UPS (it is kinda like a bomb after all) unattended an leaving your home for 6 months at a time?
Always UPS everything. But also always have a simple backdoor. I generally have 1 little desktop like a NUC running some basic Win10 OS and an install of remote software like TeamViewer. It is connected to my hardware router right after the ISP router and a backup connection. Used to be LTE everywhere, now I’m half and half on Starlink. It is then also connected to the router ports needed for management but inactive.
If I have to, remote into the NUC over Starlink. I can then reboot my main ISP box. I can eventually get into my router and enable those ports which are pre-plugged in. From there I can then access all the stack management and all the IPMI ports like iLo. It’s a virtual interface through a virtual interface. It is slow, and painful. But it works.
And it works 99.99% of the time. But even then, I’ve had to do a call of shame and walk one of my friends through which button to press as I’m on the other side of the world. In my case it was also power related but the UPS I had decided to overheat. In reality over the summer, the temps were high. But also it is a super awesome double conversion UPS. The line voltage into the UPS was dropped to below standards from the utility because their grid was overworked with everyone’s AC’s. So the UPS saw this as a line failure, kicked in the double conversion and ran happily. But it did not count as a power failure, so none of my services scaled back. Essentially it was delivering 3KW of juice from the wall through a double conversion making the whole thing super hot. Eventually it shut down for safety automatically, just pulled the plug. My NUC is on a separate little backup along with the modems and an auto transfer switch which did its thing. But there was no way to press the reset button on the UPS for a critical safety shutdown like that. It had to be in person.