I have HA running on a Raspberry Pi 4 and it runs quite stable since one year on a USB SSD.
However, at exactly 1:00 plus/minus 10 minutes each night, some of my entities (Warema awnings intgrated via MQTT) become unavailable, which I suspect comes from a crash of my MQTT broker (Mosquitto AddOn).
I saw that my CPU, RAM and temperature as well as network usage spike up around the time of the problem, like this:
How could I check if Mosquitto crashed during this time?
HA runs normally during the crash, as seen because all my other sensors (Zigbee, ESPHome, WiZ plugs, openWB) stay available and give normal values.
Check the HASS logbook for anytthing happening at 01:00:10 as that sounds like a daily scheduled systemd / cron job.
The other test is to power cycle everything and see if the issue moves in time - MQTT data can time-out (depending on retained and availability settings).
If this helps, this post!
PS The next question after Why? is ALWAYS What is in the logs?. The only question is how much help you need to interpret the logs.
I also suspect that, however I’m not the network administrator nor do I have access to the networking equipment, I just have an Ethernet port in my room where my HA and my PC get their Internet connection from.
How could I check if it really is a scheduled reboot of something?
this post!