On Tue, 30 May 2017 11:57:00 -0400, Pietro Paolini wrote: > Hi everybody, > > I am trying to configure my systemd to run my service *after* the > timesyncd service is synchronized with the NTP server, unfortunately > adding the "After=time-sycd.target" does not work as I can't see any > change in the behavior - my service is started immediately anyway. > > Reading here https://github.com/systemd/systemd/issues/5097 it looks > like there is a problem with that target but I am not sure where to > start from, is it feasible to patch the systemd-timesyncd daemon to > correct the behavior ?
Yes, this is a problem with timesyncd, and it should eventually be patched. The problem with patching it is that without re-thinking something, it is impossible to have it block time-sync.target without also risking systemd-timesyncd.service from timing out. The only solution that I can think of is to have a second helper process that can't time out that waits for a signal from timesyncd; and have that process block time-sync.target. But, the systemd team hasn't yet decided if this solution is acceptable to them. So, anyway, I went ahead wrote such a helper program, and a wrapper for timesyncd that handles sending the signal. > As alternative I could write my wrapper around the daemon as done in > the link reporting the issue, both ways I will need to read > something enlightening on the matter, would you be able to point me > into the right direction ? You should be able to install and use the wrapper that I wrote. There's nothing specific to my use-case about it. If you install it, then the "time-sync.target" should "just work" correctly with timesyncd. Of course, in the long-term it would be preferable to patch timesyncd, instead of having to install separate wrapper to fix the behavior. But for today, it works. -- Happy hacking, ~ Luke Shumaker _______________________________________________ systemd-devel mailing list systemd-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/systemd-devel
