My meshed C-3 and C-7 were only getting reboots at updates until very early this year when something changed and uptime dropped to about 4 to 5 days. I migrated Z-devices and rules from C-7 to a C-8, leaving LAN and cloud stuff on the C-7 to "split the load", and my two bulbs on the C-3. Now the C-8 is slowing after 4 days.
About a month ago, out of frustration, I started deleting apps and devices, one by one. Withings, SmartHQ, Hub Info, HPM, Magic Home lights, Open Weather Map, IFTTT, Combined Presence, Harmony Hub, Smart Bathroom Fan, Maker Api, Dashboards, Thermostat Controller, Room Lighting, Alexa, CoCo Hue. I even rewrote rules to avoid using the changed trigger but was still losing 120+ MB/day of freeMemory. I quit sharing those two bulbs from the C-3 and shut down Hub Mesh and VOILA.
I've now added everything back to the C-8, leaving the C-7 empty and added a few rules and devices to the C-3 so I can run the two hubs in parallel instead of meshing them and things look much better. I went 6 days before the last update and still had 430 MB on the C-8.
Going back through release notes, the only Hub Mesh change that matches the timeframe of my problems beginning is the change from UDP to TCP. I don't know enough about network communications to know if there is something about my network that may be a contributing factor or not. Usage stats on the hubs have always been low. I've never seen an elevated load warning and communication between meshed hubs was always very fast.
So, in summary, the eventual sharing of only two bulbs via Hub Mesh was too much for my setup, and I don't know why.
