I was in the same boat as you were. I wanted to have a Grafana dashboard with an influxDB, I wanted to have homebridge running as that is what my wife is used to and use my Echo to let my house talk to me. The only thing I don't do is Chromecast, which is also known to have issues. I was dealing with daily reboots myself and can happily say that I am without now for at least a month and a half!
here is what I did:
InfluxDB logging: instead of using the on-system Influx app, I deployed a Node-Red instance that listens to the eventsocket and pushes the events to an InfluxDB. You can find more information here:
Homebridge: I was also using original on-system Homebridge app. I completely rewrote the homebridge plugin to use the MakerAPI. You can find more information here:
For my Echo, I use Alexa TTS Manager written by @ogiewon. More information here:
Lately, I started to deploy a multi-hub strategy, which has shown to run very reliable. I use HubConnect to link them together. I have one hub with radios enabled (Zwave and Zigbee) and one hub with LAN devices and custom code. I am currently in the process of splitting up my zwave and zigbee hubs into two hubs though. You can find more about HubConnect here:
I don't think this step is needed but my desire to tinker makes me buy more and more hubs......
Please keep in mind, devices can also havoc your system. I had one particular Osram bulb that was defective and spammed the entire Zigbee stack to a grinding halt
