On Thu, Nov 28, 2024 at 10:42:38AM +0100, Jürgen Schönwälder wrote: > On Fri, Nov 22, 2024 at 08:06:57AM +0000, maqiufang (A) wrote: > > Hi, Juergen, > > > > > > > > I agree that if system configuration changes, any reference to it might > > cause <intended> to become invalid. As Kent suggested, how about we add > > some following text in section 6.1 to clarify: > > > > Server MUST ensure that any updates of <system> do not render <intended> > > invalid. However, any mechanism for handling these circumstances is outside > > the scope of this document. > > > > For me, this translates into "we designed something that has a problem > for which we have no solution". I do not think this is sound engineering. > > > > > Or even go one step further, although out of scope, we can gives some > > examples to show how to ensure <intended> remains valid: > > > > · Servers internally marked any invalid configuration in <running> > > caused due to <system> changes as “inactive” > > Well, we have no inactive, so this translates into "we designed > something that has a problem and the solution is to use something we > still have to define". I do not think this is sound engineering. > > > · Servers migrate system configuration update in <running>, e.g., > > by updating configuration data that references stale system nodes > > > > · What else? > > > > There were reasons why NMDA has system configuration feeding into > operational. We may just rediscover the reasons for this... >
Another option that comes to mind is that the system datastore does not feed into anything. Instead, it just exposes the (current) system configuration and if someone wants to use it, that someone can copy it into running and then takes responsibility for it. /js -- Jürgen Schönwälder Constructor University Bremen gGmbH Phone: +49 421 200 3587 Campus Ring 1 | 28759 Bremen | Germany _______________________________________________ netmod mailing list -- [email protected] To unsubscribe send an email to [email protected]
