On Mon, Apr 22, 2013 at 10:38:51AM +0200, Pavel Březina wrote: > On 04/17/2013 09:03 PM, Jakub Hrozek wrote: > >[PATCH 1/5] dyndns: new option dyndns_refresh_interval > >This new options adds the possibility of updating the DNS entries > >periodically regardless if they have changed or not. This feature > >will be useful mainly in AD environments where the Windows clients > >periodically update their DNS records. > > > >There is one place (in IPA dyndns code in this patch but also in AD code > >later on) that I wanted to discuss specifically. It may happen that the > >periodic update would trigger going online in which case the online > >callback would fire and another dyndns update would be invoked as an > >online callback. To prevent a race between these two updates, there is > >an interval, currently hardcoded to 60 seconds that would just make the > >next update quit without doing anything. Ideas on how to fix the problem > >without a hardcoded timeout are welcome. > > I haven't gone through the code yet, but I encountered similar > problem in sudo. There is a boolean in sudo_ctx that says whether a > full refresh is already in progress or not. If it is then the online > callback just returns. > > See sdap_sudo_full_refresh_online_cb().
Yes, that's what I modeled the update after, but the sudo periodic task does several things I don't really need. Is there anything in my patch that is not correct? _______________________________________________ sssd-devel mailing list sssd-devel@lists.fedorahosted.org https://lists.fedorahosted.org/mailman/listinfo/sssd-devel
