Tell HN: Recent SUSE update caused my first real dataloss in 20 years
I cried actual tears of furious rage today at whoever thought this: https://github.com/systemd/systemd/blob/main/src/shared/clean-ipc.c was a good idea.
I routinely treat /dev/shm as a "that will persist until this machine loses power or is rebooted." Apparently, on April 13th, one or more twits dealing with systemd decided that I no longer get to treat /dev/shm that way.
I am trying my best to remain calm as I write this, but the aforementioned code deleted not only my primary operating directories, but ALL Of my locally attached backups as well.
In all, 24TB of BADLY deleted "rm -rf" but without deleting the directories was done, in about 21 minutes before I realised what was happening.
I now have empty directories of all of my client working directories, all of my multimedia drives that our family uses at night, all of the original writes of the births of my children... all because someone decided that removing POSIX shared memory directories when I logout belongs to systemd and not me.
FAR OUT man.
Fortunately, I have a nightly that does backups to an external machine, but it is REMOTE, and now has to do a FULL COPY of the entire thing. I am typing this from my missus' lappy while it presumably will take many days to restore.
Case end point: I only really lost today, tomorrow and maybe the next day worth of billable productivity as a result, but ONLY because I have an external nightly backup running. Local backups (which mirror all local content) were all compromised because I mount them the same way inside /dev/shm.
openSUSE/SLES/SUSE itself is not to blame, the change is a systemd change and all of the major systemd distros will do the same thing to you.
Caveat emptor.