I recently switched my setup to include a ssd which is connected via USB 3.0 to my pi. The system has been absolutely miles better in terms of response times.
Unfortunately, I’ve also started to have problems.
Every few days, not on any obvious schedule, my HA instance cannot be connected to, and I have to restart the device by cycling the power.
I haven’t done a ton of debugging in HA OS so I was wondering if someone could point me in the right direction to look at logs when this occurs, so that I can figure out what the problem is.
The above mentioned power supply is a likely culprit. But if not, it may not be hardware: things like this are also commonly caused by software, for instance memory related.
If you are on HAOS, then the supervisor integration has (disabled) cpu and memory sensors on addons. enable them and make a dashboard with cpu and memory graphs. Be sure to also put the core cpu and memory in. Then look to see of memory or cpu increases over time. You might be able to spot for instance an addon eating all memory (For instance the Alexa Media Player had a huge problem past few days. That would have shown up in core memory only probably).
In the screenshot below Studio Code was overactive because I had many tabs open there:
In addition to the power supply which would be my first thing to check. In the distant past I had an SSD connected to a pi via usb. It would sometime crash, turned out to be the cheap usb adaptor I was using having a bit too much fee play on the usb connector, a quick squeeze on the metal shell had it sorted.
I feel like the power is a real possibility here. I’ve also been struggling with my usb zigbee dongle (sky connect) recently, so I’m wondering if its just that I’m starving the device.
Well, I ordered that weird adapter, and no dice. It wouldn’t come up at all on HA. Not sure that it would require a reimage, considering it’s the same SSD.
I’m trying a different power supply. This is making me want a NUC, lol.
Sorry to hear that. It certainly could be that the new adapter on the drive is a completely new device to the PI and needs reimaged.
Unfortunate!
I am running on an old business class HP Proliant DL360 Gen7. I bought mine for $260 in 2018. $180 on Amazon now. Certainly, there are better more recent versions for similar money.
Yes, it is big, heavy, and loud. But, have a closet in the office that has all of my server stuff and it is not a bother.