This Amuses Me!
Ayer, A. (2016, September 28). How to Crash Systemd in One Tweet. Retrieved October 3, 2016, from https://www.agwa.name/blog/post/how_to_crash_systemd_in_one_tweet
From what we're covering in Continuity of Operations class, this makes systemd look dangerous in way too easy ways.
AJ Jordan likes this.
AJ Jordan shared this.
I keep hoping that somebody will do to systemd what git did to Monotone and/or BitKeeper: take the good parts and package them up into a small, self-contained thing that other things can be built on.
There's a lot of things in systemd that make it much nicer to build on than sysvinit (most recent example: systemd-cat pipes a process' stdout to the system log, but unlike /bin/logger, it takes the command-to-run as command-line arguments, so I don't need the security hole of invoking a shell or manually set up a pipeline), but then there's weird glitchy corner cases, like the article you link to, or "if somebody bounces the dbus daemon, you have to hard-reboot the machine to get things working again".
There's a lot of things in systemd that make it much nicer to build on than sysvinit (most recent example: systemd-cat pipes a process' stdout to the system log, but unlike /bin/logger, it takes the command-to-run as command-line arguments, so I don't need the security hole of invoking a shell or manually set up a pipeline), but then there's weird glitchy corner cases, like the article you link to, or "if somebody bounces the dbus daemon, you have to hard-reboot the machine to get things working again".