I want to test that the whole system backup corresponds to the original.
It would be great to understand some basic tests about rapid quality assurance because I do not have time to always .untar it completely.
Do backup backup.tar.gz
of the system with exclusions
# http://unix.stackexchange.com/a/291720/16920
sudo nice tar czf /media/masi/ntfsDisc/backup.tar.gz --exclude=/home \
--exclude=/media --exclude=/dev \
--exclude=/mnt --exclude=/sys \
--exclude=/run --exclude=/proc /
Tests by Case Studies
How to Handle cases of changed files during the backup?
tar: Removing leading `/' from member names tar: /home/masi/.gnupg/S.gpg-agent: socket ignored tar: /home/masi/.config/chromium/Default: file changed as we read it tar: /home/masi/.config/chromium: file changed as we read it tar: /home/masi: file changed as we read it
How to handle ignored sockets?
tar: Removing leading `/' from member names tar: /tmp/.X11-unix/X0: socket ignored tar: /tmp/.ICE-unix/1666: socket ignored tar: /tmp/qtsingleapplication-7af9-3e8: socket ignored tar: /tmp/.org.chromium.Chromium.2qSs7o/SingletonSocket: socket ignored
Methods to tests
- Take a md5-map of directories. Compare it to the current system. It should not have changed much.
- If md5 has changed more than the significance factor
- Some specific searches of the contents in the directory, which should be enough vertical to each other to satisfy the system testing.
- Aim: real-time quality-assurance. To distribute computing and balance between power and time-duration so
nice
. TODO timing of the process and sub-processes.
Test requirements
- Permissions/owners unchanged between backup and local files.
System: Ubuntu 16.04
Filesystem of System: ext4
Filesystem of external HD: NTFS
/run
and/sys
are also ephemeral/virtual filesystems - see LinuxFilesystemTreeOverview - and should also be excluded – steeldriver Jun 17 '16 at 18:22/proc
– steeldriver Jun 17 '16 at 19:04