I have 50 devices, 14 of them routers
I don’t know how people manages to have that many routers. I have no place to put that many routers!
My question is not about zigbee specifically, but all the things in HA that give me trouble. I have to govee lights that are zigbee and BT. They work intermittently too. They were on matter, and they were terrible. At least when I moved them to the official integration they started to work a bit mor reliably.
Yah, agree. That is probably a sadly thing I will have to do. I had one tuya device overloading my mesh with 1 message per second or more. I had to turn it off, and I have a second one that is not as problematic, but I may need to also get rid of it.
Ok, that is an interesting and easy thing to try. I will have a look. Sad thing is that I don’t have that many options to reposition the antenna. Moving the mini-pc is currently not an option.
Thank you,I will have a look at it
Wow, that is pretty impressive. That is what I want for my home, not for me, but for my family in case anything happens to me.
It was dropping, but that was not causing trouble, no?
I have a couple of z-wave devices that I’m not using because the only controller I have is a little rpi hat that sucks. What are the advantages of z-wave? Is there any controller that has it all?
Troubleshooting means that I enter in a hall and the light doesn’t turn on. Then I have to bring out my phone and check the motion sensor in HA. The motion sensor is detecting presence. Okay, then the problem will be the automation. Oh no, the motion sensor has been detecting presence since yesterday, but still zb2mqtt says it is working, but in reality it is not. That means go to the zb2m dashboard, check the state, see that everything reports to be working (no, that is false info), and I have to reboot the motion sensor, for example.
Another example of troubleshooting is:
- Getting into the kitchen and again no lights turned on.
- Check the two presence mwave sensor and see that none of them is reporting presence. Or worse, one is stuck in presence detected for 2 hours and the other works fine, but because they are in a group, the group doesn’t work properly
- Check the light sensor, just in case, it works fine.
- Plug and unplug both of them and wait for them to go up again and see that they still don’t report presence.
- Have to go and reboot Home Assistant.
- Wait for it to start.
- The problem persists.
- Then go and reboot the VM.
- After 3 minutes of messing around, things start to work again.
As a first step, why don’t you stop using the Apps (add-ons). Put Zigbee2mqtt in an LXC, so that if you need to restart it you aren’t restarting the Home Assistant VM.
You can use Proxmox VE Helper-Scripts to install most of the Home Asssistant Apps (add-ons).
I doubt this will bring any benefits to the table. A restart of a vm is not something that takes a long time. The issue here is more likely a weak Zigbee mesh with a coordinator placed less ideal. 14 routers, 35 end devices and a competing IKEA Zigbee network is not ideal.
I agree that this is the issue, but by putting Zigbee2mqtt in it’s own container you can restart it without taking down the rest of Home Assistant. A VM may not take long to reboot, but it does stop everything for that period of time.
Since the OP’s current solution is to ‘reboot’, it would be easy to add a cron job to reboot that Zigbee2mqtt container periodically. I know it is not ideal, but is cheaper than adding several routers.
Go crazy with Energy Monitoring plugs, you are adding routers whilst getting the benefit of collecting additional data.
Well yes, but that is the workaround - if I had the problem I would rather try fixing the underlying issue (weak mesh and poorly placed coordinator?) than start a whole migration process to reboot some seconds faster.
That’s your problem right there:
An LQI of 28 is abysmally low and would absolutely lead to the issues you’re seeing. Please follow the guide linked earlier on how to improve your zigbee mesh, because you either have massive interference, or else your routers need to be moved to more strategic places, or both of the above.
LOL, I was already wondering why I don’t see the LQI…
But…already found out why I don’t see it…
The card only appears when there is at least one device with LQI < 50.
When All devices are ≥ 50 → the card is hidden on purpose
1 Like
Philips Hue bulbs and their other lights (and plugs ofc) are excellent routers for example. They´re little bit pricey compared to Tuya trash but definitely worth it.
Restarting a VM may not take that much time, but restarting home assistant definitely takes. It easily takes one minute after the VM is up, which can take up to, I don’t know, thirty seconds.
That said, I have previously had my Zigbee controller attached to a Raspberry Pi, properly positioned (on my previous house), and I have to say that it was a nightmare to deal with. I’m not sure I want to go back to having a separate system just for Zigbee. Having to do updates required me to go to a separate place. Having to debug problems required me to go to a separate place. When something failed, I had to look into Home Assistant first then into the raspberry pi. I was not happy about it.
The thing is those take some space, and I don’t have that much space left behind in the things I may want to monitor additionally. On top of that, I think I have enough “switches” positioned not too far from each other, and other zigbee router things.
THis is how my network looks like:
(if there is a better way to export it other than screenshot, please let me know)
EDIT:
O, here is a better export of the map:
My point was that looking at startup times and how many seconds / minutes you waste every time you need to reboot is probably the wrong way to attack the problem. You want to figure out why you are forced to reboot every month, not shorten down the boot time 
The routing table error is probably due to a weak mesh. From your picture it seems that you have poor meshing overall.
I’ll try to post a little map of where my devices are and see if people can help me figure out why my mesh is so poor, because I don’t think they are that much apart from each other.
As a side note, can someone recommend me a good PoE controller?
There one that has Node-RED integrated. Maybe I can use that one so I can have a little bit more reliable automations that don’t depend on some assistant in order to work?
I think the combination of the router/end device ratio and that the coordinator is placed isolated from most of the network is what you want to look at. You could try moving the server and coordinator temporary more centrally (if that is an option) to see if that helps on the mesh.
PoE coordinators: https://smlight.tech/
I would skip the ones with all the bells and whistles, if you already have a proxmox host you should run apps like NR there.
Are the dual ones worth it?
- Zigbee + Thread / Matter / Z-Wave (model-dependent)
For example? to have both Zigbee and Z-wave?
I know I can run those apps (like zb2mqtt, node-red) in proxmox, but if I can have a reliable little dedicated device… it feels tempting
Historically multi-radio has not been the best success, but things might have changed. I use two SLZB-06p7, one on Thread and one Zigbee.
I have a slzb-mr3 in the office on poe, it is configured for both zigbee and thread (but not actually using the latter).
It performs very well, i was amazed by the range.
1 Like
I forgot to mention that I have Loxone on my home. I paid the installer for a fully wired setup but they messed it up and now I have several Loxone Air devices. I’m not sure in which frequency they operate and if they can cause interferences