Do you see more than one temperature sensor listed in the UI graph? Does the TrueNAS UI allow you to specify which sensors to monitor and graph? I know with Core it does not. This is why for some NVMe’s, you will get alerts for exceeding a “critical temp” that is tripped by Sensor 2 instead of Sensor 1. Sadly, you can only see Sensor 1 in the graph.
If not, check with nvme:
nvme smart-log /dev/nvme0 | grep Temperature\ Sensor
Run a full scrub and repeat the command while the SSD pool is scrubbing. See which sensor increases more than the other.
You can use watch to have it automatically update every second:
watch -n 1 "nvme smart-log /dev/nvme0 | grep Temperature\ Sensor"
My guess is that I/O activity on the SSD pool is causing Sensor 2 to exceed 60C, but the UI only shows you the temperature of Sensor 1.
I am of the belief that all NVMe drives in your NAS server should be installed with a heat spreader. They’re cheap and easy to install[1] and make a significant difference in temperatures, especially under load.
If you’re feeling bold, ask the moderators to re-open this feature request.
The reason each one comes with multiple thermal pads is because you’re supposed to stack them on each other if using only one is too thin and does not make proper contact between the heatsink and NVMe. While you don’t want it to be extremely tight and hard to screw in the top plate, you also don’t want an air gap between the pad and top plate. ↩︎