@willfrazer what are you trying to flash here ? If it's not proprietary that would be useful for us to test.

@aethernet Unfortunately, it is proprietary. The description above is the closest that defines what the image contains. What I can say is that we experienced the issue in 3 different geographies on completely different systems. The md5sum of the file was always inconsistent with etcher but reliable with the other tools mentioned above. We have recommended etcher for some time as it is simple enough for non-Linux & non-technical people to use, that's the main reason we would like it fixed - and it has worked reliably in the past. This latest version lost me several days work assuming the problem was in something that I did. I only realised it was etcher when I used the most recent version

@aethernet I meant to say that should a proposed resolution to the issue be identified, I am happy to test any interim builds against the image that has been previously failing for me.

@aethernet I meant to say that should a proposed resolution to the issue be identified, I am happy to test any interim builds against the image that has been previously failing for me.

Try what I suggested in #3942 (comment)

You can also try v1.12.1 or later from https://github.com/balena-io/etcher/releases and disable the setting "Trim ext{2,3,4} partitions before writing (raw images only)".

@net147 #3942 (comment) worked OK. We will test v1.12.1 shortly. Thanks for your suggestions.

Thanks, @net147.
I was going to propose something similar.