On Mon, 03 Dec 2018, Kamil Jońca wrote: > >> Dec 03 00:23:54 alfa systemd[1]: anacron.service: State 'stop-sigterm' > >> timed out. Killing. > > See, someone or some script told systemd to stop anacron (or maybe to > > stop-and-start/restart it). THIS is causing the undesired behavior. > > Dec 03 00:22:24 alfa systemd[1]: Stopping Run anacron jobs... > Dec 03 00:22:24 alfa anacron[8919]: Received SIGUSR1 > Dec 03 00:23:54 alfa systemd[1]: anacron.service: State 'stop-sigterm' timed > out. Killing.
Looks like something asked systemd to send sigusr1 to anacron, which is "clean exit" for anacron. Since the anacron package told systemd to use "killmode mixed", it waits for a while, then proceeds to SIGKILL everything in the group because they took too long to die. It is a bug, but on anacron, not systemd. systemd is doing exactly what it was (mis)configured to do by the anacron package. And something *is* asking anacron to restart or to stop. Look at your log rotation scripts (and also anacron's). > I have NO scripts which kill anacron. IMO it is systemd thing. What happens when the anacron service is stopped/restarted under systemd is different from non-systemd anacron, but that's because the anacron package misconfigured systemd, so it would be a "systemd thing as (mis)configured by the package". But something is triggering that service stop/restart/force-reload (can't tell which from the logs), and it is not systemd. systemd is obeying orders from some other system component. -- Henrique Holschuh
