Any chance the device is attempting to join via a router that is more than a couple of hops from the hub?
You'd expect a join to work from anywhere in the network (within range of a repeater), but after a number of reported issues with joins failing when the joining device happened to be more than two hops from the coordinator/trust center, SiLabs came up with several workarounds (changes to the way the hub's firmware is configured) to mitigate the issue...
If the joins succeed closer to the hub, it's possible that issue might be in play here: Why aren't nodes sometimes able to join (no TrustCenterJoinHandler callback) -- caused by an address resolution issue (if the node sending the Update Device command is a non-neighbor of the TC and not tracked in any of its neighbor/child/binding/address tables for address resolution)
The situation has not changed from all my posts to date. The main 2 devices used throughout this thread are the
JASCO light switch:
endpointId: 01
application: 16
firmwareMT: 1124-0054-00000003
manufacturer: Jasco Products
model: 43084
softwareBuild: 00000003
And
Sengled color light:
endpointId: 01
application: 26
firmwareMT: 1160-80CC-00000026
manufacturer: sengled
model: E21-N1EA
softwareBuild: 00000026
The general behavior is that I can pair the devices and see via the sniffer that they initially connect usually through 1 sometimes 2 repeaters. After a day or so Hubitat starts failing to send commands to 6 or more of the around 28 devices. If I wait another day some of those disconnected ones will start working again and others, that were working, will have stopped. If I use the sniffer to check I see habitat never sends a command to the devices that stop responding and (so far) for each of those that remain operational I usually see a direct link from Hubitat to the device (no repeaters being used even though the initial pairing often used one).
In general the initial pairings tend to use other JASCO switches as the repeaters of choice however I do have dedicated repeaters (about 6) throughout the house:
endpointId: 01
application: 03
manufacturer: eWeLink
model: SA-003-Zigbee
powerCluster: none
Occasionally I see them being used at pairing time.
@mike.maxwell@Tony
Boy do I feel like a dummy. I fired up a spare C5 hub and tried to pair and it wouldn't... Humm
The problem turns out to be operator error. I was plugging the outlet in while holding the button until the light flashes. As I just discovered I was supposed to plug it in while holding the button but releasing it as soon as the light came on. Jeez louise all this time screwing around with this thing. I guess I should write it down next time? Sorry folks
You're not the first to fall down this path, so don't worry too much about it. So many devices, so many different reset/pairing steps. It would have been nice if the the Zigbee and Z-Wave manuf had seen the shared benefit to making it easy by having consistent steps. Oh well...
Thanks for the details. As mentioned in our private message, your Zigbee radio stopped going offline a week ago, likely due to our engineers' efforts. If you had incidents more recently, we would like to further investigate. While the problem may be seen in compatible devices, the issue may be the result of non compatible devices. If you could provide additional details about your mix of devices, either here or in the private message that would probably help us narrow down where the issue may be.
I had 3 Centralite 4200C's lock up today. It started with one. I was pretty sure I could get it going again by unplugging/replugging it, but I had a cat on my lap and wanted to experiment. In the course of my experimentation, nothing made it go active again, and I made two other plugs go dormant as well.
I try to put the platform version being run on every post. The drop out issue is still occurring. If it is an ‘incompatible’ device then it was fine with C7 but became incompatible for C8. Unfortunately the JASCO switches (which make up the major part of my automations) are hard wired throughout the house, put in 20 months ago when I started with the Hubitat C7. One thing that has improved with the Hubitat updates is the ability for the Sengled lamps to pair, this was very troublesome in early C8 releases.
Not sure what information you need, Hubitat reports:
JASCO light switch (around 20 devices):
endpointId: 01
application: 16
firmwareMT: 1124-0054-00000003
manufacturer: Jasco Products
model: 43084
softwareBuild: 00000003
Sengled color light (around 4 devices):
endpointId: 01
application: 26
firmwareMT: 1160-80CC-00000026
manufacturer: sengled
model: E21-N1EA
softwareBuild: 00000026
Repeaters (around 6 devices)
endpointId: 01
application: 03
manufacturer: eWeLink
model: SA-003-Zigbee
powerCluster: none
If you need more details I will try and get it.
The situation has stayed the same for me throughout the Hubitat C8 releases. I can re-pair all devices and they work for a while. When paired most devices use 1 repeater hop. After a day some start to become inoperative. When viewed with the Zigbee sniffer I can see no attempt by Hubitat to send commands to those ‘dropped’ devices. I also note a tendency for the still working devices to be direct connect to the Hubitat hub and not using the repeater path they initially paired with. If I test a few hours later sometimes a previously dropped device will start working again (usually direct connect to Hubitat, no repeater used in comms) and a different previously working device will become inoperative. As mentioned before, when a device pairs it tends to use other JASCO light switches as repeaters in its path, this seems normal since there are so many of them evenly spaced around the house.
What does seem odd is that Hubitat is sending commands to the furthest device from it without using any repeater hops (even though the radio is on low power). It seems that Hubitat has a big problem keeping the repeater paths alive and tends to fall back to direct connect. Although Zigbee is meant to be a dynamic self-healing mesh I really wish that (in my case) the system just remembered the paths at joining time and used them.
A couple of months ago when the in-laws visited I transferred 10 of my most automated lights to Home Assistant Zigbee and redirected the Hubitat rules to talk to the resultant Home Assistant devices. This setup worked solidly for a week (no drop outs) after which I transferred them back to Hubitat.
Thanks for the details. What other mains powered devices (besides Jasco switches) do you have connected to your hub (make and model).
That is possible. We have seen issues with some Zigbee 3.0 devices that have not been tested by our engineers and that seem to run fine on C-7 but have problems on C-8 - E.G. Aqara, (some) Tuya , Xiaomi, (some) Sonoff.
So far this device has worked each time it is needed however it is not a daily requirement.
As mentioned above there are 6 Zigbee repeaters, nothing plugged into them. They are part of a repeater group that allows me to turn them on and off en-block, they all seem to work.
Interesting, most of the other mains powered devices have been transferred to z-wave sockets. All other Zigbee devices are battery powered sensors or push buttons.
If you have enough Jasco devices like you said, I would remove the S31s from your environment. We have had reports that they are problematic on C-8 hubs.
Thanks...same here, went from 16 to 8 to 12 to 20 and then down to 4. I'm going to swap to a new C8 hub on Thursday to see if that helps.
But while I'm waiting for Thursday (wife will be gone all day so I can blow things up w/out causing more WAF problems) I'm playing around w/a different Zigbee antenna. I thought it'd be interesting to experiment w/a spare antenna I had around (had come in a set of 2). They are dual mode Zigbee/Z-Wave like the stock C8 antennas. I had been using one of them on my C8 for Z-Wave for weeks now and it improved my Z-Wave results over the stock antenna on the C8, so decided to try the second one of the pair on my current C8 for Zigbee.
Started w/the new antenna yesterday late morning at power level 4 and things were a little better but some odd behavior (devices fully controllable from their Device pages would fail to respond to On or Off commands in an automation) so I upped the power to 8 last night. Since then things have been surprisingly stable/functional. I had one Aqara contact sensor fall off last night, re-joined it and it seems to be sticking, and everything else looks OK at this point. Also just had an instance of a light not turning off when the automation sends the off command...yet fully controllable from the Device page.
The changes are too recent to make any judgements, but so far no Zigbee reboots since yesterday morning, haven't had any other devices drop off the mesh, and automations all seem to be working. I'll let it run like this for the next couple days and see how it behaves.
The one pointing upwards is Z-Wave, the one pointing downwards is Zigbee.
Switched out the Sonoff power socket for eWeLink one. Re-paired dropped devices, a day later 8 had stopped responding because Hubitat no longer sent commands to them as usual.**
Newer platform release seems to have trouble re-pairing Jasco devices, had to reboot base to get one to re-pair, still am unable to get one other to pair despite repeated factory resets and trying the alternate Zigbee pairing options (avoid version 3 repeaters and keys in the clear). Will keep trying.
I assume you removed the devices that won't pair from the hub as well (red "Remove" button on Device page). I find that Zigbee reset/re-pair on the C8 often requires that I fully remove the device for before it will re-pair to the hub. Reset/re-pair w/the original instance of the device still on the hub (where it will report previous device found) will either fail, or complete but the device still won't report/function properly.
@harold.min I did the same thing. I had an old C-7 and just reinstalled it on the network. I still have Zigbee devices dropping whenever they seem to want to. Nothing seems to get them back on the system short of Removing, resetting, and re-pairing. Then they last a few hours, days, weeks (anyone's guess), then they drop again.
I know it's not for everyone, but I've had no problems since going repeater-less for my battery devices.
Everything else is Z-Wave.
I've got some of the latest Sengled plugs, but haven't deployed them.
I don't really need to, but might try some as an experiment...at some point.
