Ralph, I had tried again with installing 18.10 as an extra O/S - twice, in fact! This seems to have disgusted the PC so much that it sent me back to grub-rescue! With the one combination of hd and partition number that found something, the config file was missing. So I did the obvious and installed 18.10 in place of 18.04, and immediately upgraded to 19.04. As I had /home on a separate partition this didn't disturb user files. A rather obvious step I guess.
I swapped the new /home with the old one and almost every thing that mattered was back. The only "clever" step I took was to change the new user and group numbers to be the same as on the old /home. I'm sure I still have some chowns to do but it's OK. Ubuntu 19.04 reports system errors about twice a day but so far these haven't disturbed anything. I lost about half a dozen applications but when I re-installed them the old config files were found and used. GIMP isn't installed with the distribution, which surprised me. > Did you have it configured to do automatic package updates? - auto > notification but not installation. I don't think an update was going > on when the problem started. My real find was that according to gparted (under the trial run of 18.10) the / partition was full. But according to "disks" it was unallocated. I increased the size with gparted but still "disks" got it wrong. Would you expect a full system partition to generate a recoverable error? > I wonder if the graphical display manager, e.g. gdm, has a recovery > mode - I can't see any reference to one online. > `ps axf' shows all processes in a forest. You might see what's taking > the time, e.g. an fsck(8). Adding a `u' to that, `auxf', adds columns > including %CPU that could show what's busy. > > `systemctl' will list the state of the system in systemd's eyes. > It tends to colour-code unusual states. > They are useful, thanks. >> I've tried several tests; the only one that showed a hard error was >> MemTest86+ when using ALL 4 CORES - it stopped at Test #7 (Block Move >> at 4096M-6144M) but with no output, the cpu seemed to have stopped. >> With only 1 core all tests were passed. > That's the kind of thing to do when all is thought well to give a > comparator. - good point. Made more relevant by discovering that this failure is well-known, but not on all PCs/motherboards Finally, the up-to-date version of flightgear, now in a Ubuntu repository, runs fine. I can invite my grandsons again. Regards, John -- Next meeting: BEC, Bournemouth, Tuesday, 2019-06-04 20:00 Check to whom you are replying Meetings, mailing list, IRC, ... http://dorset.lug.org.uk/ New thread, don't hijack: mailto:dorset@mailman.lug.org.uk