I've been hunting those memory leaks for months now 
2 Likes
I would also like to "pile on" by saying that the latest iterations, (2.2.8.x), seem to have the least amount of "memory leaks". I don't know if everything is out, but it sure seems from my graphs that there is very little, if any, memory being lost on my hubs, in day to day operations.
In my humble opinion, while there may still be bugs, it appears that they are few and far between.
1 Like
An additional clue. The hub has locked up at exactly midnight local time both nights since the soft reset. The lockup is hard, port 8081 is inaccessible, ICMP ping is unresponsive.
This is my experience as well. I have caused the memory to decline on my hub to the point that a reboot is necessary about once a week. The thing that caused it was enabling Apple’s Adaptive Lighting through Homebridge. It seems to me (and Hub Stats) that Zigbee rgbw lights may consume a lot of resources when sending thousands of commands per day to them. These are directly connected lights, not Hue, which I enabled Adaptive Lighting on to match my Hue lights. I like it enough to put up with it, and can disable it if necessary. My other hub never needs rebooted, and has 2 Hue hubs connected with CoCoHue, most of the rules, and 309 total devices connected; 68 of which are Zigbee. The hub that needs rebooted weekly only has 199 total devices and very few automations.
