I realize this is anecdotal evidence, but this is what my results are.
7 hubs, all connected to HomeKit instances. 4 go to one home, 2 go to another, and 1 goes to a third. All Hubitats go into the Hub Load is Severe mode, and rebooting them fixes the issue for 12-48 hours - usually in 24 hours. Some have several devices shared with HomeKit, some are very minimal. I disabled the app on 6 of the hubs, leaving one enabled. The one I left enabled has 3 devices shared. After rebooting them all, I waited about 36 hours. 1 of the hubs shows the Severe Load message, the other 6 do not. Guess which one has the message? Yes, the one with the devices still connected to the HomeKit instance. Possibly a random result, but I doubt it.
I have the same issue. Hub load severe message after ~24 hours. App and Device runtime statistics doesn’t seem to show anything amiss. A reboot clears the message, but then it reappears. This behavior started after I started using the integrated HomeKit integration. Any ideas?
Interesting. I moved most of my HomeKit Integration devices back to homebridge but left a smattering of 9 devices on the native integration and am experiencing the severe loads. I might try and move the rest back to homebridge and see if I still get these issues.
My hub load is back to "elevated." I assume it will be back to "severe" sometime in the next few hours as we approach 24 hours from my last reboot. This only started happening after I enabled the integrated HomeKit (an unofficial diagnosis).
FWIW: Since updating to 2.3.4.138 I haven't seen "load is severe" messages, but now I get a non-responsive hub. I can't login to the hub which forces me to power cycle the hub. That clears the issue until the hub eventually becomes unresponsive again. Lather, rinse, repeat...
After a reboot, when I check the "past logs" I see smart home device activity in the log during the "non-responsive" time as if the hub was operating. But it wasn't - my home devices were not responding which is what alerted me to try to login to the hub and discover I couldn't.
Just to weigh in, I am seeing the same thing on 139. Devices go “no response” and hub alerts elevated and then severe. Reboot fixes it for a day or so.
Every time I peek at the logs, load on the Homekit app is usually the highest busy%, but never more than 1% of total load.
No devices are showing high load either.
Feel’s like there is a problem somewhere not visible to user logs.
Also, since this started I have also had corrupt database events 3 times in the past 2 weeks. Prior to this starting I had corrupt database events zero times in over a year.
Maybe coincidence. Maybe not coincidence.
Hard part is there is nothing suspicious in any of the logs other than the elevated and severe load alerts.
I’m probably reverting to Homebridge V2 if this keeps up more than another week or so.
Similar experience for me. I'v only had to do a database restore once. Still having to occasionally login in via the :8081 port with my MAC address to get it to reboot as the normal login screen will go non-responsive when the hub gets put into this weird state.
