Update on my experience:
Previously, my Pi would crash once or twice a day with the white disk-activity light flashing furiously when installed in the DeskPi Pro case. I have also posted my experience (also as leicray) on another forum hoping for guidance from there too: DeskPi GitHub site It might be useful to read through that thread to understand the sequence of events and to see the experiences of other users.
The best advice on this forum was to keep trying with the Pi in the DeskPi Pro case and wait for a crash and then take a look at the contents of the kernel log just before the crash:
/var/log/kern.log in the hope that might reveal something useful.
Before we get to that, it took nearly three days this time before it crashed again. Luckily, I had a a terminal window open on the desktop this time. However, some commands worked within the terminal, but others did not. For example I was able to
cat the contents of kern.log but could not save it, as
nano would not load and redirecting to output from
cat to a file would not work either.
At this point, I could not connect to the pi using ssh or VNC. In addition, the desktop was frozen apart from the open terminal. It’s probably best to say that the pi was only just alive at this point.
The content of kern.log shortly before the “crash” is:
pi@raspberrypi:~ $ uptime
13:12:00 up 2 days, 21:16, 2 users, load average: 13.02, 10.45, 9.56
pi@raspberrypi:~ $ cat /var/log/kern.log
May 3 11:46:09 raspberrypi kernel: [244220.207312] xhci_hcd 0000:01:00.0: WARN Cannot submit Set TR Deq Ptr
May 3 11:46:09 raspberrypi kernel: [244220.207325] xhci_hcd 0000:01:00.0: A Set TR Deq Ptr command is pending.
May 3 11:46:09 raspberrypi kernel: [244220.358262] usb 2-1: reset SuperSpeed Gen 1 USB device number 2 using xhci_hcd
There are ten virtual bonus points on offer for anybody who can guess how I did capture the content of kern.log and it was not by copying manually from the screen to a plain text file.
The good news is that the UK supplier (PiHut) has issued an RMA number and provided a pre-paid return label. In an ideal world, I would like to have a working case, but there is clearly something very wrong even with the revised board.
Any additional comments or encouragement would be most gratefully accepted.