I've been one of the few, special people w/long-term Zigbee issues on my C8. However, recently the hub appears to be finally settling down and Zigbee is behaving well. (Cue fireworks.)
As of the past few weeks Zigbee devices are no longer regularly falling off my C8, and it has not been having Zigbee radio reboots for quite a while before then. Zigbee-based motion & contact automations in RL and RM are running as expected, and even new mmWave Zigbee devices are playing nicely w/the team.
Changes I've been making:
Starting some weeks back, rather than re-pairing devices that fell off the C8 back to the C8, I began to move those devices to my C7. The devices stayed and played nicely on the C7 (which is meshed to the C8). That process, along with removing some Zigbee devices I wasn't actively using reduced Zigbee devices on the C8 from 120 to 93. The C7 now has 38 Zigbee devices.
Devices that consistently fell off the C8 were ones that had been rock-stable on the C7 that I migrated from. Devices that regularly fell off the C8 included many of my Samsung/Centralite leak sensors. I also would typically lose Iris v2/v3 Motion sensors, and Visonice MCT-340 contact sensors. In some cases the Iris and Visonic would refuse to re-pair w/the C8. Aqara contacts would fall off now and again, but overall were my most reliable Zigbee devices.
Repeaters on the C8 include: SonOff USB Dongles, the tiny Tuya USB repeaters, and plugs (Iris, Centralite, Sengled, Innr, Ikea). No Zigbee switches, dimmers, or in-wall outlets on the C8, and mmWaves are on the C7.
Other significant changes:
Moved C8 from Zigbee channel 20 to 15 a few weeks ago to avoid a neighbor who had decided to plop his Wi-Fi across Zigbee channel 20. The C7 was already on Zigbee 25.
Also, I got a new tinfoil hat
C8 is still doing all of my automations and all but one LAN integration. Over time I'm going to look into moving some of the automation/LAN integrations load to the C7 as well to reduce the work the C8 is handling.
Things are feel more solid than they ever have, which is nice. This does feel like things are finally going to be OK again. Most importantly, WAF is finally on rebound. That is a big deal.
Well, @danabw: good for you! It's about time that you got aboard the train with everybody else!
What I have always found puzzling in these situations is trying to really establish root causes - because if we get the real root cause, then of course the situation could be properly dealt with.
Of course, all of those devices that you were having trouble with (Centralite leak sensors, Visonic MCT-340 Contact sensors, & Iris V3 Motion Sensors), I've got on a C8 with no issues. That's the way these things always go - it's always mysterious and unknown as to why these strange things happen.
So, what does your intuition tell you was that real root cause?
Hmm, those have been some of the better behaved devices on my hub.
Yep, I don't get why some have such horrid issues with a device, and others no problems at all. It must be down to a particular device firmware, or combination of devices on a hub?
The latter is the most likely, LOL. That's my wife's opinion, anyway.
I've seen such a measurable/obvious improvement since I started putting the "troublesome" devices that fell of the C8 on the C7, that I have to give a nod to that change. Not other action I've ever taken resulted in the the same obvious and rapid improvement in Zigbee stability.
So maybe a combo of 1) Too many Zigbee devices and 2) Some of the devices that I moved to the C7 were not behaving well on the C8 in some unknown way. Still...I (and HE staff checking my detailed logs) didn't see anything from my devices that was obviously an issue, so that does leave me thinking just the reduction in the number of devices was the most important change.
Of course, there are folks w/many less devices than me who also have had issues, so this does end up being a significant head-scratcher.
Exactly, on my C7 they were always perfect, zero issues. And I still have a fair number of the same devices on my C8 humming along happily. Just hard to figure.
But I'm not one to question good luck, I'll take it!
That the C8 took over direct communication to an incrementally large number of devices (due to improved radio & antenna capabilities) such that it no longer had repeaters helping to handle (and/or slowing) the traffic? Or might dialed up radio power of the C8 itself contribute to stepping on too many simultaneous attempts to reach it?
Counterintuitive given we usually consider "cutting out the intermediary hop" an improvement, ...but I'll throw the idea into the discussion to be counted or dis-counted.
I do seem to recall that several people claimed to have better results with some devices (even on the c4/5/7) when they were βforcedβ to jump through a repeater to get to the hub.
And some people with C-7s are seeing problems too.
But in any event, serious on your having got things working reliably with your setup.
TBH one thing that would be really useful in troubleshooting this would be a few extra diagnostics - my very limited understanding is that it's a Zigbee radio issue with multiple potential causes, but that additional data on the cause of the reset is available via the existing radio firmware. If that data could be passed back to the HE UI so there were clues on whether the underlying cause was... [waves hand vaguely] too much traffic, iffy content, weak signals, whatever... that could be very helpful. If nothing else we could pass more info back to Hubitat themselves so they had a better view of the big picture.
@danabw@jtmpush18 curious if you upgraded the firmware of your Centralite/Iris devices with a ST or Iris hub prior to switching to HE years ago. Wondering if firmware version plays into the situation of liking the C8 or not.
Before Iris went under years ago, @ogiewon mentioned he bought an Iris hub on sale so he could upgrade firmware on all his Iris devices and I followed suit and did that as well. Iβve been holding out on migrating to my C8 given the various Zigbee issues folks have been posting about.
Interesting question. I did not do any "conscious" FW updates on Samsung/Iris/Centralite stuff while on ST, and I don't think ST will do it all on its own, IIRC. So I should be on the shipping version(s) of FW on all of those devices AFAIK. A quick check of Iris v2 motions and Samsung leak sensors shows:
Iris v2 motions - all on below FW, both those still on C8, and those moved to C7:
Leak Sensors - oddly/interestingly, the Samsung leak sensors moved to the C7 all have the newer FW, 14 vs. 11 on the ones left on the C8:
Fell off and moved to C7 - FW 14
Still on C8
So go figure...of course no idea if this is correlation or causal, but it is pretty interesting that the Samsung leak sensors w/newer FW were the ones that kept falling off and I moved to the C7.
Maybe I'm dense (or might just need another coffee) but I'm not sure what to conclude from your post..
What I initially took from the topic headline: C-8 Zigbee reliability has improved... issues resolved.
But here's what I think you've actually said: You started with a stable Zigbee setup on C-7... when you moved essentially the same Zigbee devices to the C-8, Zigbee went from 'working well' to unreliable and maintenance prone. Fixing the situation required (in addition to trying a different channel) moving Zigbee devices from C-8 back to C-7, with more offloading of C-8 Zigbee still to come. Without hub mesh and moving devices to C-7's Zigbee network, there would be no improvement.
Correct me if I'm wrong, but as I understand it from other communications on this topic, your stable C-8 Zigbee mesh now consists of zero directly connected end devices (all the end devices in your C-8 mesh are now child devices of other neighbor Zigbee routers)... and you had already tried multiple C-8 hubs to rule out bad hardware . Firmware updates and swapping out C-8 hardware wouldn't fix periodic C-8 Zigbee 'offline/reboot/online, issues and dropped C-8 child devices... but removing devices from the C-8 mesh did.
I might have misconstrued your post, but I'm trying to square the "finally settling down and Zigbee is behaving well" comment when it appears that nothing Zigbee related in the C-8 has improved; you've just reduced its Zigbee device complement to something it can handle reliably.
Is that right?
Maybe my title needs to be re-worded? I think the details are clear from the OP - that I've gotten an overall mesh improvement that was mostly accomplished (I think)
by moving a subset of Zigbee devices from the C8 to the C7. At the current device level on the C8 (93) it is now reliable. I had tried multiple C8 hubs. The third one was the best in terms of stability, but on its own w/all my devices it didn't solve all my issues.
But yes, in a world where I only have one C8, I was not able to find a change that resolved my Zigbee reliability problems.
I do still have a lot of devices connecting directly to the C8 (if the HE Zigbee graph isn't lying).
Yes, exactly, and I thought I said that pretty clearly in the OP, laying out the key changes I made to get there, especially including the before/after number of devices on the C8. If things had just magically started working w/out doing anything on the C8 then the OP would have been a lot shorter. Hopefully that clears things up.