pure ubuntu with firefox, chrome and chromium-browser and basic linux tools. Ubuntu Desktop 20.04 (MATE) Dockerfile with NoMachine 7.1 remote access. I had to keep 10 snapshots a day for the devs there b/c they were using SVN (blech!). docker-ubuntu-mate-desktop-nomachine 7.1. IMO, best practice for critical data is to do incremental snapshots with rsync hard linking (save space).

Something about 'clean up after yourselves' and 'please use the head nodes for launching software' just didn't get through to them. I think maybe I didn't really care about that job - they ran their /home mounts at 99% utilization 99% of the time and preferred to SSH into nodes in the HPC cluster instead of using a job scheduler like grown ups do. I wasn't going to be responsible for project data in upwards of 8PB. They began utilizing USB 3 drives to backup projects for themselves. I loved the look on peoples' faces when they asked me to recover 300GB of data b/c they didn't pay attention to the autofs prefix when doing stuff. When I worked for a seismic processing company, we used XFS with the 'noikeep' mount option. This is the benefit of using ZFS, LVM, or any other filesystem that allows for snapshotting. We use a ZFS filesystem at my new gig, which allows for snapshots, so at the very least we roll back to an earlier snapshot or just recover the missing data from the snapshots.

At the very worst, you'd lose a day of data. This way you can nuke the root fs at whim, reinstall, and restore.

In the future, it's always best to know the locations of data files and have backups of /home, /var/lib/mysql/data (and/or dynamically changing data).

Nomachine ubuntu 20.04