So, third another time I had to update all the automations, scenes and references because another tuya power plug became unreliable: it sometimes works, it sometimes ignores what I ask it to do, even using the physical button on the device.
So I decided that this was enough common of a problem to apply a better solution.
Usually, when I repurposed a device, or replaced it with a better one I had to do a lot of updates in a lot of places, all from memory and by inspecting/searching the yaml files, not fun. Sadly HA does not provide you with clues of where a disabled device is mentioned, so this is all manual work.
In contrast, the management of my kitchen has been way easier. There, everything is in a group because everything is redundant: 2 lights, 2 motion sensors, 2 light sensors. And guess what? every-time I had to replace a motion sensor (because shelly devices are not reliable either) it was as easy as go to the group, add the new one, remove the old one.
This experience lead me to take the decision to use groups for everything, even if it is just a single device, so I can just reference the group and never have to update anything when I have to replace devices. This also gives me the ability to give those groups sensible names, without compromising the devices with names that will become obsolete (how many Bedroom TVs plugs do you have now in your office monitor?).
Is this advisable? Is it a bad decision? Will I face problems Ia m not able to anticipate now?
Apart of avoiding using the device itself in your automation, as per @tom_l’s message, you should also take a look at the Watchman, which helps you finding those entities that are not existing enymire but still used in your automations or settings. You get the file and line number of each entry if those.
Honest question. How is that better than using groups?
I didn’t knew that entity_ids can be re-assigned to a new device. Will you get an error if you try to assign a device_id that already exists?
I activated watchman very quickly, but sadly I don’t have much time to properly configure it, and I have the impression that I am not making the most out of it.
Is there an easy way to add it to a dashboard and just check it there? What I have now requires me to get a message in my telegram bot, and I even think I have to trigger it myself.
Sorry, I meant entity_id. Let’s say I add the new one before deleting the old one. Will I get an error when trying to set the device_id? Or will I get undefined behaviour?
You can use it without setting up any of the parameters. Maybe you should even try it first without change the standard settings and then later find which settings your should change.
90% of time I look for the report on my files. I’ve never used the notification for Watchdog.
You can also find an example of a card on the documentation here. It’s easy.
And as it can be triggered by a service call, you can easily add to your automations to have the report updated daily, or have a button to update it in your dashboard.
I see advantages and disadvantages on each one. The main advantage of going to the groups approach is to have icons handled when you change the device (and have a new entity_id), and not having to rebuild linked things as you mentioned.
However, groups don’t have the same flexibility with icons like entities have, so the user experience is better using entities everywhere.
And, create a group for every single entity sounds way more work than renaming a couple of entities when I replace a device. I don’t know about you, but I don’t replace my devices every day, so that is not a hard work.
You can assign a previously used entity_id to a new entity as soon that is not in use anymore.
Yes, you will get an error message if you try to assign an entity_id currently in use by another entity.
I was using groups but switched to just updating “entity_id” and haven’t looked back. The main impetus for the change was not being able to group sensors, and I didn’t want a different solution for them.
It also simplifies things - I send myself warnings when devices go offline for a period, and getting them twice (device + group) was annoying.
The same here. But in my case not because we lack groups for sensor and not because I don’t like to have double notification of entities unavailable.
My single reason is because it’s much easier to rename a few entity_id fr8k time to time than maintain a million groups.
You know, I was almost sure that I installed watchman, but when I went to HACs I didn’t saw anything in my integrations, so I just installed again.
Then, for some random reason today I decided to take a look at my folders, and some ethereal force made me look at app-daemon, and there it was, watchman. I installed watchman as an app-daemon app some time ago.
Is the “integration” the new and recommended way of using it? I think that just for the already available configuration card it’s probably worth it, but I would love to hear your opinion.