Hi, I need some help debugging why my hub frequently freezes and becomes unreachable on my network.
The hub crashes seemingly randomly, sometimes staying up for a few weeks and other times just days. When it's crashed, the device has the green light on but does not respond over the network, not even the diagnostic page. The only thing I can do is pull the power to reboot, then it works for a random period of time until it crashes again.
I have removed all automations and custom code/apps, and the only thing I have active are a few z-wave light/fan switches and the google integration.
Things I have tried:
Updating hub firmware. Currently on 2.3.4.130, but it's been problematic since at least 2.3.3.121.
Update router firmware. Tried Netgear firmware and FreshTomato.
Soft reset and restore from backups
Swap out different ethernet cables, usb cables, and power supplies
Connect to different ethernet ports: on the router and going through a switch
Connect to different power sources: UPS battery backup, UPS surge only, and wall
Set the hub's Networking "Ethernet speed" setting to fixed 100mbps
Set the hub's Networking "Auto reconnect" setting to 300s
Holding down the hidden reset button on the bottom of the hub when it's crashed. The hub's LED color stays green.
Change DHCP lease time to infinite
Current setup:
Netgear R7000 router running FreshTomato 2023.2
Hubitat C-5 hub on 2.3.4.130
Hub directly connected to the router's ethernet ports
Hub setup with a DHCP reservation
Hub's "Ethernet speed" set to auto
Hub's "Auto reconnect" set to 300s
I'm completely out of ideas and would love to figure this out.
Thanks!
Before pulling power, try to unplug the ethernet connection and plug back in. Mine does something similar but can't pin it down to anything and its so random I gave up hunting it down. But I can unplug the ethernet and plug in and its all back to normal for some random undetermined amount of time
The hub's ethernet speed has been set back auto for the past few crashes.
I've checked both my routers (one in AP mode) and they're not set to use jumbo frames. None of my computers have it on either. Is there a suggested way to find if any devices are sending large frames over the network?
I've also tried manually reconnecting the ethernet cable while it was crashed yesterday and it didn't seem to help. Though I suspect something like that (rebooting my router) did bring a dead hub back to life once, but I haven't been able to reproduce that fix. Though I'd hardly call that a viable long-term fix.
I've tried connecting the same power adapter to three different power sources and it's crashed on all of them. I've also tried a different power adapter on those same sources and it didn't stop the crashing.
The LED always stays green when it's dead. Holding down the reset button did not reboot the hub and stayed green. I know that button is for resetting the network IP settings, but the documentation states it would also reboot the device so I gave it a shot.
It sure does, though I didn't verify if the behavior is different for the different hub revisions.
... turn the hub over and use the point of a pen to press the small button just beneath the surface of the hub casing for at least 7 seconds.
The LED on the front of the hub will blink, then the hub will reboot (at which point the LED will turn blue as part of the normal boot process before again turning green once the hub has fully booted).
Hmm - try a soft reset with a database restore. Corrupted databases are also known to cause the hub to hang.
When the hub's LED is green, and it is inaccessible, do automations still run? Meaning if you activate a motion sensor or change the status of a contact sensor, does a z-wave/zigbee switch turn on/off?
Maybe it's your router DHCP Lease Time. If it's on the set no time limit or turn it off for your Hubitat. I did have this problem with my C5 a while back.
Though my previous message about it dying after renewing IP could have been misconstrued as it dying immediately after the renewal - I just don't have any logs (in my router or the hub) beyond that point to know exactly when it went offline.
Oh, another important data point to mention is that when the hub dies like this, it even fails to perform its nightly backup. That makes me think that the whole hub is frozen, not just its network interface.
That's a good data point. My hub usually run out of memory when I don't have network connection after a day or two because I have quite a few apps needed network connection. I am guessing because of the amount of logs generated. Doesn't mean this is also your case. Just throwing it out there and hoping it could help.