MoCA is honestly really great. I’ve been running 4 MoCA nodes (2.0 and 2.5) in my house for 10 years now, over pretty shitty RG58 cabling and a bunch of splitters, quite long runs too, along with 1 gig DOCSIS 3.1 modem sharing the same cabling, and it has been absolutely bulletproof and reliable (especially compared to the powerline ethernet I was using before, which was absolutely terrible). Plenty of bandwidth for full duplex 1 gig - really the only downside compared to 1 gig ethernet is a few extra ms of latency.

Running ethernet cables everywhere would be very expensive in my situation, and while I considered it a few times to get more bandwidth between my main computer and the NAS which sits in my homelab in the basement, I still find it hard to justify the cost. If I were to pull the trigger now, I think I would go with fiber (or maybe both) - but I would have to replace so much stuff to get up to 10 gigs, all my homelab equipment is 1 gig only, server nics, NAS nics, switches, etc… it would not be cheap, beyond even just the cost of running the cables. So I am just sitting back and waiting to see how the technology evolves for now.

What kind of battery did you put in ? A standard alkaline battery has different voltage discharge curve compared to the recommended rechargeable NiMH batteries (example)

You can’t really measure the capacity of a battery directly, all you can measure externally is the voltage, so the logic has to make some assumptions about the type of battery being used when calculating percentage. Since they recommend the LADDA rechargeable batteries, which are NiMH, they probably set the 100% threshold lower - perhaps as low as 1.2V, so as long as whatever battery you are using is above this voltage, it will show up as 100% - it might take longer for an Alkaline battery compared to NiMH rechargeable, but later you might have the opposite problem, when the reported percentage will start dropping faster for Alkaline vs NiMH, due to the different discharge curve characteristics.

More sophisticated devices can sometimes have a learning function, where they observe the voltage curve over a full discharge cycle of a battery, and then adjust the % reporting to more closely match, but since every battery can be different, this is still not always accurate. The discharge curve can be affected by temperature as well, further complicating the issue.

Since these sensors are supposed to last up to 15 months on single charge, and it actually wouldn’t be too unusual for a device like this to report 100% battery for up to the first 1/3 of its battery usable life (especially if using Alkaline batteries), 3 months is probably still within reason. There is often some intentional padding as well by manufacturers, as they don’t want customers to see big drop in battery percentage early on, as this might cause people to make ultimately incorrect predictions about total battery life, and complain or return the product.

Have been using Linux since Slackware in the early 90s, both for work and at home - been through a lot of distros since then.

Currently use Debian on my servers and VMs, and CachyOS (with KDE Plasma) on my main desktop. I like the bleeding edge of Arch for my desktop, and find CachyOS to be a very sensible plug and play way to get into Arch (I tried to run Arch before, and also EndeavourOS, although that was a while back and not the smoothest experience for me - things might be more mature now, but CachyOS was the one that finally reached the point for me that allowed a full switch from Windows on the desktop to Linux)

Most of my servers run virtualized on Proxmox, which itself is Debian based though. I also run pfSense as my firewall/router which is FreeBSD based.

Funny enough, the origin of that character is 姦 - which goes all the way back to the earliest forms of Chinese writing (the bronze inscriptions) and was a pictogram of 3 women together (see here), which represented “evil, treacherous, scheming, traitor, adulterer” and similar. 奸 was a later variant form with same meaning (the 干 was added as a phonetic component, and 3 女 were reduced to just 1), which was later adopted as the simplified form of traditional 姦.

+1 for MoCA

I switched from powerline to MoCA about 10 years ago, and it was a huge step up. Even though it’s half duplex, since MoCA version 2.5, there is enough total bandwidth available to sustain 1 Gbps in 2 directions simultaneously, so it is functionally almost equivalent to full duplex 1 gig Ethernet (except for few ms of extra latency)

pihole often doesn’t help, as many IoT devices either use their own DNS servers and ignore the one provided by your network, and sometimes even skip DNS completely and just connect to hardcoded IPs directly. Even blocking DNS at the firewall/router is getting more difficult with increasing use of DNS over HTTPS and custom DNS server IPs that aren’t in public lists. (I block all known DNS server IPs at my firewall, forcing any device to use my own DNS servers, but even that is not always completely effective)

It’s usually best to isolate IoT devices on VLANs with no internet access (blocked at the router/firewall) Although there are now even devices that can autonomously connect to external WiFi networks like Amazon Sidewalk, to gain internet access and bypassing any restrictions you might try to place on them…