Comment 36 for bug 1863672

First of all, I'm sorry I appear to have been neglecting this bug. It's hard to debug something you can't see! And I've had lots of other things on my plate unfortunately.

I don't think anyone can claim with a straight face that casper-rw is anything other than opaque. We could have an argument about whether 'writable' is better or not but that's not very interesting.

And, well, I still can't reproduce this. I have tried, several times, on real hardware. It's a bit harder to test on hw currently because my usual test machine is in an office I can't travel to thanks to the COVID-19 lockdown, but I tried your instructions and booted the resulting disk on my machine and it still mounted the casper-rw partition.

So let's try something else. I've uploaded a hacked up initrd to https://people.canonical.com/~mwh/casper-initrd.gz. If you can drop this over the one that came on the ISO and boot there should be two files present: /run/udev-ls.txt and /run/udev-dbg.txt. Can you boot a failing system and attach the files to the bug? udev-dbg.txt will be a few hundred kilobytes, the other one will be small.