Re: [systemd-devel] Antw: Re: Q: Implementing logrotate's postrotate with systemd
Am Di., 11. Juni 2019 um 16:18 Uhr schrieb Reindl Harald : > > > > Am 11.06.19 um 15:00 schrieb Ulrich Windl: > Reindl Harald schrieb am 11.06.2019 um 14:30 in > > Nachricht <917331d8-845f-54d5-908c-e6c7d124a...@thelounge.net>: > >> > >> Am 11.06.19 um 13:34 schrieb Ulrich Windl: > >>> I have a forking service (with a PID file) that can reopen the logfile > >>> after > >> receiving SIGHUP. In the past I had implemented "rc{service} rotate" to > >> send > >> SIGHUP to the daemon as "postrotate" action. After converting (actually > >> being > >> converted ;-)) to systemd I dropped the LSB script, and wonder which > >> command > >> to use as "postrotate" action: > >>> > >>> Should I implement a oneshot service (using "systemctl start {service}") > >> that does depend on the actual service and send a SIGHUP on start, or is > >> there a more elegent solution? > >> > >> that's what reload is all about > >> > >> [harry@srv-rhsoft:/etc/systemd/system]$ cat named.service | grep Reload > >> ExecReload=/usr/bin/kill -HUP $MAINPID > > > > The manual page says it's about "configuration reload". I was talking about > > logfile rotation (my service does not suport configuration reload (other > > than restart)) > > frankly it's nothing new or uncommon that services close and re-open > their logfiles by "reload" and that's really not systemd specific, > sysvinit had reload too > > it's you turn what happens with "systemctl reload yourservice" becaus > ethere is no default action for that unless "ExecReload" is specified Of course you can do whatever you please in ExecReload, but as said I think it's good practice if "reload" has a certain consistent meaning across services so admins not familiar with a particular service know what to expect from "systemctl reload". See also https://github.com/rsyslog/rsyslog/commit/fd26a42bdc04eaf497cafd9ef806a54f3de1a7e9#diff-c64e6ed7f40ecd1530e093a77c9465f6 -- Why is it that all of the instruments seeking intelligent life in the universe are pointed away from Earth? ___ systemd-devel mailing list systemd-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/systemd-devel
Re: [systemd-devel] Antw: Re: Q: Implementing logrotate's postrotate with systemd
Am 11.06.19 um 15:00 schrieb Ulrich Windl: Reindl Harald schrieb am 11.06.2019 um 14:30 in > Nachricht <917331d8-845f-54d5-908c-e6c7d124a...@thelounge.net>: >> >> Am 11.06.19 um 13:34 schrieb Ulrich Windl: >>> I have a forking service (with a PID file) that can reopen the logfile >>> after >> receiving SIGHUP. In the past I had implemented "rc{service} rotate" to send >> SIGHUP to the daemon as "postrotate" action. After converting (actually >> being >> converted ;-)) to systemd I dropped the LSB script, and wonder which command >> to use as "postrotate" action: >>> >>> Should I implement a oneshot service (using "systemctl start {service}") >> that does depend on the actual service and send a SIGHUP on start, or is >> there a more elegent solution? >> >> that's what reload is all about >> >> [harry@srv-rhsoft:/etc/systemd/system]$ cat named.service | grep Reload >> ExecReload=/usr/bin/kill -HUP $MAINPID > > The manual page says it's about "configuration reload". I was talking about > logfile rotation (my service does not suport configuration reload (other than > restart)) frankly it's nothing new or uncommon that services close and re-open their logfiles by "reload" and that's really not systemd specific, sysvinit had reload too it's you turn what happens with "systemctl reload yourservice" becaus ethere is no default action for that unless "ExecReload" is specified ___ systemd-devel mailing list systemd-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/systemd-devel
Re: [systemd-devel] Antw: Re: Q: Implementing logrotate's postrotate with systemd
Am Di., 11. Juni 2019 um 15:00 Uhr schrieb Ulrich Windl : > > >>> Reindl Harald schrieb am 11.06.2019 um 14:30 in > Nachricht <917331d8-845f-54d5-908c-e6c7d124a...@thelounge.net>: > > > > > Am 11.06.19 um 13:34 schrieb Ulrich Windl: > >> I have a forking service (with a PID file) that can reopen the logfile > >> after > > receiving SIGHUP. In the past I had implemented "rc{service} rotate" to send > > SIGHUP to the daemon as "postrotate" action. After converting (actually > > being > > converted ;-)) to systemd I dropped the LSB script, and wonder which command > > to use as "postrotate" action: > >> > >> Should I implement a oneshot service (using "systemctl start {service}") > > that does depend on the actual service and send a SIGHUP on start, or is > > there a more elegent solution? > > > > that's what reload is all about > > > > [harry@srv-rhsoft:/etc/systemd/system]$ cat named.service | grep Reload > > ExecReload=/usr/bin/kill -HUP $MAINPID > > The manual page says it's about "configuration reload". I was talking about > logfile rotation (my service does not suport configuration reload (other than > restart)). Right, I wouldn't use ExecReload= for this use case myself. If I run "systemctl reload $service", I expect that the service reloads its configuration and I think it's good practice to be consistent in that matter and not overload ExecReload= with "rotate log files". -- Why is it that all of the instruments seeking intelligent life in the universe are pointed away from Earth? ___ systemd-devel mailing list systemd-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/systemd-devel
[systemd-devel] Antw: Re: Q: Implementing logrotate's postrotate with systemd
>>> Reindl Harald schrieb am 11.06.2019 um 14:30 in Nachricht <917331d8-845f-54d5-908c-e6c7d124a...@thelounge.net>: > > Am 11.06.19 um 13:34 schrieb Ulrich Windl: >> I have a forking service (with a PID file) that can reopen the logfile after > receiving SIGHUP. In the past I had implemented "rc{service} rotate" to send > SIGHUP to the daemon as "postrotate" action. After converting (actually being > converted ;-)) to systemd I dropped the LSB script, and wonder which command > to use as "postrotate" action: >> >> Should I implement a oneshot service (using "systemctl start {service}") > that does depend on the actual service and send a SIGHUP on start, or is > there a more elegent solution? > > that's what reload is all about > > [harry@srv-rhsoft:/etc/systemd/system]$ cat named.service | grep Reload > ExecReload=/usr/bin/kill -HUP $MAINPID The manual page says it's about "configuration reload". I was talking about logfile rotation (my service does not suport configuration reload (other than restart)). > > [harry@srv-rhsoft:/etc/systemd/system]$ cat rsyslog.service | grep Reload > ExecReload=/usr/bin/kill -HUP $MAINPID > ___ > systemd-devel mailing list > systemd-devel@lists.freedesktop.org > https://lists.freedesktop.org/mailman/listinfo/systemd-devel ___ systemd-devel mailing list systemd-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/systemd-devel
[systemd-devel] Antw: Re: Q: Implementing logrotate's postrotate with systemd
>>> Michael Biebl schrieb am 11.06.2019 um 14:13 in Nachricht : > A separate oneshot service sounds like overkill. I would probably use > something like > `systemctl kill -s HUP ${service}.service` > If your sevices spawns multiple processes and you only want to send > SIGHUP to the main process, you should add a `--kill-who=main` Thanks! I wasn't aware of systemctl's "kill". Regards, Ulrich > > All documented nicely in > https://www.freedesktop.org/software/systemd/man/systemctl.html > > Am Di., 11. Juni 2019 um 13:34 Uhr schrieb Ulrich Windl > : >> >> Hi! >> >> I have a forking service (with a PID file) that can reopen the logfile after > receiving SIGHUP. In the past I had implemented "rc{service} rotate" to send > SIGHUP to the daemon as "postrotate" action. After converting (actually being > converted ;-)) to systemd I dropped the LSB script, and wonder which command > to use as "postrotate" action: >> >> Should I implement a oneshot service (using "systemctl start {service}") > that does depend on the actual service and send a SIGHUP on start, or is > there a more elegent solution? >> >> Regards, >> Ulrich >> >> >> >> ___ >> systemd-devel mailing list >> systemd-devel@lists.freedesktop.org >> https://lists.freedesktop.org/mailman/listinfo/systemd-devel > > > > -- > Why is it that all of the instruments seeking intelligent life in the > universe are pointed away from Earth? > ___ > systemd-devel mailing list > systemd-devel@lists.freedesktop.org > https://lists.freedesktop.org/mailman/listinfo/systemd-devel ___ systemd-devel mailing list systemd-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/systemd-devel