Thank you so much for sharing your design thinking. I have no comments but will be trying your concepts. I am considering whether it might overlap on Gridfinity bins for organising things.

I think you’ve got the key point here. Most of the commentary on AI focuses on its impact on people that are already good at their jobs. Where I see it having most impact is for people who aren’t good or aren’t capable. It doesn’t make them a good coder, but it might make them a bit better than before.

The same is true for professional and technical writing. There may be an explosion of people ‘delving’ into topics, littered with em-dashes, and it might take them a while to produce it, but the principle time saving is that I’m not reviewing and editing all the garbage the first time round.

Rebooted a bunch of bits and it seems to have sped up again. Weird really. I love zigbee for the low cost and low power consumption,M but as a tech nerd I hate that I don’t have the same tools as I would have with an IP device.

I switched out the router and things started working. Very weird, but I can’t tho jot anything other than it being the Virgin Media hub not liking it. Apparently they have history on this.

Yes, everything looks right. I moved dhcp resolution from the router to technitium recently, but hadn’t set up local resolution.

I’m currently thinking the router is the culprit. Here in the UK there are lots of forum posts complaining about the Virgin Media gear. Nothing specifically describes my problem but I’m going to try a new router over the weekend.

Thanks for the suggestion, I’ll grab a cap to check.

I’m running tcpdump -i any port 53. I can see the outbound request but not the reply. Will the cap show me anything more?

Thanks for giving it some thought!

I have been testing using dig @192.168.0.249 study.lan

3, 4, and 5 work for TXT, NS and SOA but doesn’t work for A records. I think this rules out a simple network issue?

Thanks for replying, I appreciate the response.

I’m running dig @192.168.0.249 study.lan from my client (a MacBook).

If I run ‘dig @192.168.0.249 study.lan TXT’ I get a correct response (I have added a txt record)

If I run ‘dig @192.168.0.249 lan SOA’ or ‘NS’ I correctly get the records for the zone.

I think this eliminates the possibility of it being a routing error?

Have you ever felt tempted to put a dashboard around the place? Maybe the rest of the family aren’t bothered about longer term information etc.

I was thinking things like weather and power usage might help my family. I also thought something that showed who was home would be cool.

I’ve got a decent number of local manual controls, but not all of them. For example, some of my wall switches operate the relay because they are just turning on and off the power. Others I have disabled the relay on because the lights themselves are WW/CW tuneable and HA controls the colour during the day.

I’m wondering about having another look at zigbee groups and commands for the simpler automations in the house. I avoided these because they aren’t really visible to HA and I didn’t like having two automation ‘languages’ at the same time.

Overall, how long do you think you could cope without your HA platform before it becomes an issue?

I was thinking about matter yesterday, I like the idea of being able to have multiple controllers. My house is half wifi devices and half zigbee. I’d been favouring zigbee recently because I don’t want to swamp my network with device packets, but maybe that needs a rethink. At the very least my wifi devices all have esp home configs that could be configured to fall back to defaults.

Some cool examples there, I’m going to think about them. I particularly like the walking ones.

I want to love dashboards. I love the idea of a control centre in each room but I just can’t get to the point of winning with them

I totally understand mission critical motivations, but I reached a different conclusion from you. I’ve been HA’ing for a long time and everything dies eventually.

Do you have a backup HA green in the cupboard? my wife would murder me if I couldn’t get the house back in 24 hours. I want to use hardware that you can buy literally anywhere so I don’t need to keep a backup.

I’m not there yet, but I havve moved to running HA on a proxmox server and have used my HA backup to recover from a software failure. I’m now thinking about what the same would look like for a hardware failure, either the mini pc or the zigbee dongle.

I like this, but I don’t like that rooms are cold unless I hang out in them. I live in an old stone house and I’d rather have a room schedule than a room presence sensor.

I’ve done something similar. I use the excellent scheduler from Niels Faber. I have room radiator TRVs. I have a helper entity that counts the number of radiators that have an open valve, and an automation that triggers when this changes to/from zero. This automation adjusts the central nest thermostat, converting it to more or less a posh switch.

My radiator valves use Better Thermostat and external temperature sensors, and the whole thing keeps my house warm in every room. I need to write a GH blog post about it, but happy to answer questions/share bits of config

Big shout out to Adaptive Lighting. Absolutely love this integration

I can’t exactly solve your problem, but when I wanted to get HA running on proxmox I used these scripts

Tteck

Completely painless and running in almost as little time as it took to download the files.

If you’ve got access to the file system I think you could remove the custom component there - can’t exhausted resources if there’s no code!

I did wonder if that was the one that did it. That’ll teach me to update multiple things at once.

The only thing I lost was a day of recorder data and since I repaired it before bed I guess if I went and recovered the previous build I’d lose what I had overnight in the switch back.

Thanks for letting me know though, I’ll warn my friends before they update.

This happened to me, I’m running HAOS on proxmox. Ended up restoring from backup. Rollback from the CLI didn’t fix it either.

I don’t know how tech savvy you are, but I’m assuming since your on lemmy it’s pretty good :)

The way we’ve solved this sort of problem in the office is by using the LLM’s JSON response, and a prompt that essentially keeps a set of JSON objects alongside the actual chat response.

In the DND example, this would be a set character sheets that get returned every response but only changed when the narrative changes them. More expensive, and needing a larger context window, but reasonably effective.