On Thu, Nov 22, 2018 at 02:15:59AM +0100, Arnt Karlsen wrote: > > > It has the potential to bring down the system, thus one could even > > > argue upon a RC severity, but they accumulate so slowly (apparently > > > two per day) and take so little resources that I wonder what could > > > have slowed down Arnt's machine so badly. I've waited past only a > > > single midnight though, so it's possible badness accumulates > > > superlinearly. > > ..nah, all it takes is setting up systemctl restart loop cronjobs to > juuust carry on looping.
systemctl restart is never executed on a sysvinit system. > > The original bug reporter has yet failed to say what's going wrong > > here. I am tired of running after poorly worded bug reports. > > ..which part of:" # daily restart of atop at midnight > 0 0 * * * root if [ -d "/run/systemd/system" ]; then systemctl \ > restart atop; else /usr/share/atop/atop.daily \& ; fi" "on a > non-systemd box" did you miss? a non-systemd box doesn't have /run/systemd/system, thus the then part of the if statement is not executed, systemctl is never executed (and would only fail since it doesn't exist on a non-systemdbox), and thus /usr/share/atop/atop.daily is executed. The only "bad" effect I see is that a cron process and a shell stay around. > > Adam, what exactly has the potential to bring down the system? > > ..the cron job _loops_, because it cannot for very good reasons > find /run/systemd/system nor systemctl, and therefore _loops_. Explain how a loop can occur. I don't see a loop on a freshly installed Debian unstable sysv system. I also don't see how a loop can occur. But I might be wrong here. > ..this bug is insisting on assuming systemctl etc systemd commands checking for /run/systemd/system is the canonical and documented way of checking whether we have systemd or not. If you don't want to have unused systemctl commands in Devuan packages, then don't take Debian packages. You're free to do the work yourself then. Greetings Marc
