Quote Originally Posted by utu View Post
I also note that persistence files can occasionally get corrupted; usually in my case, from imperfect
shut-downs. It helps to have a pristine backup of persistence files, just in case.
I find it useful to occasionally examine what dmesg has to say in this regard:
Code:
dmesg | grep -i corrupt
This will not only find problems with persistence, it will uncover similar problems
which may have crept into Windows and data partitions.

An additional technique I use is to make cloop overlays of well-developed persistence files
as additional fat32-partition-KNOPPIX[n] files which are pretty much immune to shut-down corruption.
After adding such an overlay, I immediately use gparted to create a new 'empty' (reiser)
persistence file for additional changes which I might make to my Knoppix LiveUSB.

I try to minimize whatever is left to additional read/write persistence. After I get my Knoppix
'immunized' somewhat in this way, I consider on-going r/w persistence as 'temporary' and just
resurrect from notes any losses due to imperfect shut-downs.
A minimalistic back-up of small r/w persistence files is an obvious alternative I might use here.
I wonder what is the reason for this. I thought is was confined to running KNOPPIX under kvm, because I first detected it there. But it is not. So I have put a small script, fix_fs, in /usr/local/bin, to check the persistence file systems. Turns out they have to be fixed most of the time nowadays. I'm pretty sure it wasn't like this before. I'm not talking about crashes or emergency halts. This happens with all standard methods of shutdown that I have tried.

Perhaps we should pursue the "read-on-startup" and "write-on-exit" method, I think some distros have adopted it.
Seems that mainstream Linux is turning into ever more of a buggy beast these days. Maybe not entirely coincidental.