No gpio but old centrino laptops make excellent low power servers. My primary server was a first gen centrino from 2011 up until recently and I think it only used 12w idle after putting a SSD in there. Had it’s own UPS built in.

I migrated from WordPress to Grav and love it. If you know markdown Grav is easy, efficient and FAST.

I paid for a year of an incogni family plan and wiped my whole household off the internet. Within a month google searching largely stopped coming up with hits other than accounts I personally created. I’ve been quite happy with the service, It took no effort just $100.

I went a little crazy and setup my own wireguard VPN network, all the remote hosts connect to the VPN and the primary server connects to each of them and pushes backups. Because I use btrfs and lots of snapshots I use btrbk, annoying to setup but now my hourly snapshots get pushed everywhere, minimal bandwidth and it flawlessly has worked for years.

I do this except the offline copies are raspberry pis, they grab an update then turn their network card off and go black for about a month. Randomly they turn on the network card, pull a fresh copy and go black again. Safe from randomware and automatic.

I have a trailer (workshop) with solar power, batteries, a raspberry pi controlling everything and a cellular hot spot. It pumps all the solar, battery information and light controls over MQTT and home assistant over cellular. So yes its possible, what do you want to do?

Shit I thought I was so damn novel. Blast you.

I’ve owned and deployed a lot of pi, every model, and in my experience when I have similar instability as you described its related to the sdcard. Either the sdcard itself or the tray soldered to the pi. I had one pi that would corrupt the sdcard without fail after 2 months and I played with bending the sdcard metal tray inward a little to help press the card better into the contacts and the problem went away. Try fiddling with the sdcard holder or different sdcards.

I host it on the host that runs the script and proxy it. I have one mission critial pi that is my uptime bot, pi hole and backup VPN if my elaborate server falls on its face. But you could easily use docker volumes too, and have the script push to that folder.

Yep, here is the yaml but redacted

- type: entities
            title: Communication
            entities:
              - type: weblink
                name: Webmail
                url: https://postale.io/
                icon: mdi:email
              - type: weblink
                name: Mattermost
                url: https://mm.stuff.com
                icon: mdi:chat
              - type: weblink
                name: Mumble Server
                url: https://mumble.stuff.com
                icon: mdi:radio-handheld

Similar, but more fancy, I have a bash script that runs every 15 minutes and ingests a config file. The config file has a super simple CSV format of every service I have. It checks that all the services are operational and generates an HTML file from it. If any services are down the HTML will show its down, otherwise its just a helpful link.

I just made a landing page in HASS, if you’re already running three instances could you make a page in one?

Older speakers like that use always on transformers, constantly wasting energy to keep the core energized. You’re correct those cannot be made any more, they must use efficient switch mode supplies.

+1 on this idea, going to toss in my recommendation for an AMD 5600g second hand. Its basically a laptop CPU with built in GPU that handles my jellyfin transcoding without issue and has a super low idle power rating if you pair it with a quality, small PSU.

Thats friggin bananas. Do you live somewhere with lots of hydro power? Your cost is less than 1/3 mine…

My SSDs use negligible power at idle, I only noticed a 1w increase when I installed two. Almost ‘free’. Also your 0.14kwh is almost certainly just the cost to generate the power minus the delivery fees. Where I live the delivery fees double my true per kWh cost. Double check your bill and divide your monthly consumption by your monthly payment to find the real cost.

Any quality brand SSD (Samsung, Kingston, WD, etc) is going to be more reliable in every way compared to mechanical disks, they just cost a lot more right now. Do NOT buy off brand, random Chinese SSD, you will regret it.

When I bought them 2 years ago power in MA was $0.46 per kWh, this included transmission costs and all the other fees. 15W cost me $4.80 a month, so $57.6 a year and $230 over 4 years. At the time 14TB mechanical disks were about $300 so it was about a $270 ‘premium’ for solid state over mechanical so I exaggerated the ROI, but to me the 2x price premium was worth it for silence and no latency on retrieving my data. So in summary the ROI for me was more like 8 years, ignoring the many advantages of SSD.

Does no one care about power consumption? Mechanical disks use, in my experience, 7-15w all day all the time just idling. If you live in a high energy cost area the ROI on going SSD can be as low as 3-4 years. If you can afford it, splurge for SSD. I spent ~$800 on two 8tb SSDs and I’m very happy with the choice.

Yes and because wiregurad is stateless you’ll need a script that checks if your DNS endpoint has updated and restart the wireguard interface so it pulls the fresh DNS/updated IP address. I had to make said bash script for my nodes.