I just reproduced the same 'failure to join' behavior; this time with the exact same C-8 I've been joining stuff to (in 'debug' mode) successfully for several days; still on same .121 firmware.
Key to producing it (at least for me, since I apparently live in beta tester hellmouth) is to simply soft reset the hub. I left the C-8 on same channel as before (14), no apps.. device joins fail in various ways (stuck on "Found a Zigbee device, initializing...... please wait, still initializing... ", "looks like trouble" etc.)
I did not use any repeater, but of course all join attempts were within a foot or so of the C-8.
EDIT: I did not restore any database, just tried joining devices after soft reset.
Logs showed only one of three trials actually produced a 'Device' instance (it was an iris V2 contact sensor), but the device didn't actually work when activated. Interestingly, the event log did show
"Initializing Zigbee Device " followed by 64-bit MAC and device ID's for all failed trials as well.
Again, this was using same C-8 with same devices previously joined without issue prior to soft reset (though some likely joined for first time with earlier releases of firmware)
Odd...I don't expect that every user experiencing these issues types of issues has done a soft reset, but likely at least some have - unfortunately maybe even more likely after problems started, as a part of troubleshooting that we might suggest to them. So we may have inadvertently reinforced the issue?
Or...the soft reset is a coincidental issue to something else we aren't/can't see happening. [sigh]
Yes very odd. Soft reset is supposed to enable fixing a configuration problem not causing one; yet clearly I just used it to turn a functional C-8 into one that will no longer perform Zigbee joins.
Don't ask me to speculate (though I'd love to; 'Resets' were a major system test function back when I worked on jurassic hardware... they're not trivial; lots of stuff is happening that sets the stage for all functions thereafter)
.
I can't remember ever doing a soft reset on the C8. Most likely never did since I migrated my zwave network from day one and thankfully that's still going strong. I did reset the Zigbee radio though.
Not likely that soft reset is the cause of this issue for (probably all) folks that have encountered join issues, but maybe it will allow putting the hardware into the same ' join failure prone' state to allow the root cause to be determined.
..anyway, blame @csteele for putting the idea in my head, lol.
Some progress reported (at this point for other devices) but encouraging to hear that HE are identifying/working through the issues, and finding those fixes should help w/figuring out others. Good to see this news...
I hear yeah man. personally I'm just thinking you guys should just roll the stack back to whatever it was. let people know to get better repeaters so devices stop dropping off. Just my thinking . I'm sure you guys will figure out. Thanks for all the efforts so far.
FWIW, I restored the database to soft-reset C-8 (.121). Reset/rejoining devices was quick and normal.
I then tried 'Remove Device' with Third Reality motion sensor; it failed to rejoin. Timed out, 'looks like trouble..'; device left with LED blinking blue.
So the only interesting part of the 'soft reset' in the scenario I described is that my troubles in joining coincided with not restoring any database after the soft reset. Join issues seem to be correlated with not having a prior instance in the database; I'm not seeing an issue resetting/rejoining devices that already are in the database. Delete them (or soft reset without a database restore) and joining is problematic.
I've deleted Hue and I believe Visonic devices and been able to add them back to the hub...several others have reported more success in joining when they deleted the device and then re-joined it. So removing a device doesn't seem to block re-joining it in those instances.
Only problem with that is for the person that just received a fresh Hub. The only way I've been able to get stubborn devices to finalize was to rebuild the Zigbee network then pair. Sometimes I would have to put the hub into pairing mode the hub finds the device then do a rebuild network while it's trying to finalize. It's crazy Weird. you'll actually see the device listed as Dev in the zigbee logs then watch it finalize and work fine.
I just paired 11 Hue outdoor Motion sensors in Home Assistant. No wonder people are having problems with them. If you don't do a factory reset even if you removed them properly from whatever coordinator. They are stubborn as .... to get paired to whatever you're trying to use in place of the original coordinator. I had to hard reset them multiple times to get them to pair fully. Zigbee2mQtt you can see a little more of the process.
When I first got the C-8 I ran into several problems with device resets and rejoins; I experienced issues which at that time involved an inability to rejoin a device that had been previously joined and reset... it would not rejoin again without first deleting it from the database (or unless it was rejoined through a router); as I recall, that was prior to version 2.3.5.110, and that particular problem was indeed fixed by firmware.
With each subsequent release through .121, all the existing device resets and rejoins I have done were for devices that already existed in my C-8's database. Until today I have not repeated the exercise of joining a device for which no current device instance exists
What I am noticing now (for the first time in several releases having run into issues joining devices) is almost the exact opposite of the pre-.110 scenario.
This appears almost as if there is something database related which is interfering with the completion of a successful join; but now it seems to happen with devices for which no current entry exists in the database.
I don't fully understand what's going on and can't generalize the behavior of my setup (which consists of very few devices) to what someone would see in a typical installation.
But I can see how it would slip under the radar for someone who's only attempt to replicate the problem involves rejoining an existing device...
So, I am one of the unlucky who has been struck by C-8 ZigBee issues after migrating from C-7. The migration SEEMED to go extremely well, but then cracks started to show on the ZigBee side (I've had no trouble with Z-Wave).
Long story short, on 121, I was still seeing issues with various devices including Hue Outdoor Motions, a Tuya 4 in 1, and a couple of my Sengled contact sensors (Of course these all worked without issue on the C-7). I finally started removing any device that had signs of failure from Hubitat, resetting them, powering them down (no idea if that made any difference), and then pairing them back in fresh. I used virtual devices and Swap Apps Device to save a huge amount of work. I also dropped ZigBee Power to 8. I've been watching for a few days and so far they have all continued to work. They DID work fine for a few days in the past, so I am reluctant to declare victory, and I still feel like the system seems fragile, but I'm hoping things at least hang in there while stability improves with further firmware tweaks.
So we may have inadvertently reinforced the issue?
