Devices stopped working in our house and upon inspection, our C8 (running 2.4.0.150) seemed to have had its Z-Wave subsystem "lock up".
When I logged into the web UI, I first noticed a "Last cloud backup failed. Failed steps: Z-Wave" notification (as well as a new platform avail - .151). When I tried to open the Z-Wave section of Settings, it was just a blank page (other than the header).
I went ahead and let it do the platform update, thinking that might work, but after rebooting on 2.4.0.151 I still was unable to use any devices, and Z-Wave was still completely dead.
I did a full power cycle (shutdown -> unplug -> wait -> replug) and everything seems ok once again.
I am just concerned this might happen while I'm not at home, and therefore unable to do anything about it.
Is this a known (recurring?) problem?
Can the firmware be updated to allow a full hard-power-cycle on the Z-Wave interface via software in case this happens again or in a situation where the hub is not easily physically reachable? I understand the desire to keep the Z-Wave subsystem "alive" during reboots to make things faster, but I feel it would be beneficial to be able to power cycle it via software in the cases like this.
I think the only real solution to this is to use a wifi plug on the Hubitat hub itself. After shutting it down, you would need to go into the app for the wifi plug on your phone and turn the power off for 30 seconds, then turn it back on.
For a full automation of this, it would need a routine in Alexa or Google that has the wifi switch connected. The Alexa/Google routine would first tell Hubitat to shutdown, by changing the state of a virtual switch, then it would wait for a bit, and then turn off the power to the hub, then wait 30s, and then turn the power back on. On Hubitat, when the virtual switch is flipped by Alexa/Google, then have it do the hub shutdown in a rule.
Yeah, I mean, I know hacky ways to "work around" this, but:
This shouldn't be happening, so I assume there is a bug somewhere, perhaps in the Z-Wave firmware?
There really should be a way to actually hard-power-cycle the Z-Wave subsystem to be able to resolve these sorts of issues remotely and without needing to use some additional 3rd party device to do a hard power cycle.
I'm unsure if there is any debugging info or troubleshooting steps I can take if I see this happen again - thus my post here in case someone from Hubitat can chime in
Well not really. I have a c8. it happened to me where the zwave system was not loading like i showed in my screen shot. i had to turn off, power cycle and while the upgrade zwave was showing prepower cycle, it is gone now. Will see of it happens again but that was not on c7.
Since it is doing a full power off at the source, I'm not sure how the hub could ever turn itself back on. I think this would require an actual hardware change on a new hub design to keep some component powered while everything else on the hub gets powered down internally by that component, so that it can turn it back on.
I agree, the need for the work-around shouldn't be needed, as Zwave should just stay working. "What broke my Zwave" is a very common topic here, I even have a post out there from when my Zwave was locking up, but I found the devices that were causing the problem and I got it fixed by replacing them. There seems to be just too many reason why ZWave can go down, though.
What I mean is, when the hardware (the entire Hubitat device) is/was being designed, there should have been some thought put into "what if this 3rd-party component locks up". Meaning - some sort of ability to "power cycle" that without needing to physically unplug and re-plug the device.
You could likely use a MOSFET (or other transistor) or even a tiny DIP relay to do this sort of thing.
Perhaps this can be considered for the C9 (or whatever)?
That said, I don't know what vendor Hubitat is using for its onboard Z-Wave controller - but they should consider reaching out to the vendor to see what they recommend in order to "power cycle" that component without having to hard power cycle the entire device. Point is: I'm sure there is a Vcc input that can be cycled on the Z-Wave section of circuitry to at least do this.
Probably not something that can be retro-fitted to existing hardware, since it seems like that was not part of the hardware design.
This current situation just really sucks for any remote deployments or other situations where hard-cycling the power is difficult.
