Request of some helpfull functionality

[tackin]

I'm using z2m for over 2 years now together with a Sonoff Dongle-E as Coordinator.
My setup is docker based on a Raspi 4 which is booting from USB-SSD.
A container for z2m, one for Mosquitto, one for Homeassistant and so on. The Raspi 4 is mostly under 5% CPU-use and also has not much traffic on its LAN.
The Coordinator is connected to a Raspi1 via 30cm-USB-extension and connects to my z2m-Container via ser2net
and LAN-cable.
Works mostly great. But ....
Last week I found some of my HA-automations failed because of missing input from zigbee-sensors.
But all Zigbee-devices are working fine. Only Z2M did not get any device-updates via LAN for some reason.

1. Looked into the z2m-log: no Errors
2. Looked int the raspi1 for power-problems (in the past) -> nothing found, runs rocksolid
3. reobooting z2m via the WebGUI of Z2M - > does not help, still no input form the devices to z2m.
4. restart of the z2m-Container -> Sucess! Everything up an running again

So, a watchdog in z2m cannot help me with that problem. And it seems to me, that this problem is not a z2m-problem at all.

I would like to install a cron-job on the host for doing some checks inside the z2m-Container and restart the container from the hostside if needed.
Therefore I would like to do something like
docker exec -it some-vodoo-api-call
Where "some-vodoo" would be: "please answer true/false if there is any devices "last-seen"-value is shorter than x seconds"

If not a single device has a "last-seen-value" below x seconds I would assume that this error within docker has occured and I need to restart the container to keep the system up and running.
This ist not able to fix via a Homeassistant-automaiton because HA cannot restart an other Container. The restart-request needs to be run from the cronjob from the host with the apropriate userrights.