Red Hat Bugzilla – Bug 812799
systemd kill and isactive commands
Last modified: 2012-05-17 11:30:34 EDT
Description of problem: After killing the service with "systemctl kill arptables_jf.service" the service is still active according to "systemctl status arptables_jf.service" and "systemctl is-active arptables_jf.service". I think the kill command should either stop the service or report some non zero exitcode to indicate unsuccessful kill. Version-Release number of selected component (if applicable): # rpm -q arptables_jf systemd arptables_jf-0.0.8-24.fc17.x86_64 systemd-44-4.fc17.x86_64 How reproducible: always Steps to Reproduce: 1. systemctl start arptables_jf.service 2. systemctl kill arptables_jf.service 3. systemctl status arptables_jf.service 4. systemctl is-active arptables_jf.service Actual results: Kill exits with zero exitcode, but does nothing. After that status says the serice is active, but is-active says it is inactive. Expected results: 1) both systemctl status arptables_jf.service and systemctl is-active arptables_jf.service should report the same 2) kill should either kill or stop the service and return zero exitcode or fail and return non-zero exitcode.
arptables_jf in rawhide now (bug #754423) provides native systemd unit file so I'll comment how that works. (In reply to comment #0) > Expected results: > 1) both > systemctl status arptables_jf.service > and > systemctl is-active arptables_jf.service > should report the same This seems to be fixed with the rawhide version, i.e both command show that the service is active. arptables.service contains RemainAfterExit=yes which means that the service shall be considered active even when all its processes exited. > 2) kill should either > kill or stop the service and return zero exitcode > or > fail and return non-zero exitcode. According to systemctl(1) 'kill' sends a signal to one or more processes of the unit. But in case of "one-shot" service like arptables there's no process to send signal to. I'd say that returning zero exit-code is ok in this case, but that's question for systemd guys. Anyway, I'm closing this BZ because (1) seems to be fixed in rawhide and (2) is ok in my opinion.