[Opensim-users] Consequences of vicinity

Stefan Andersson stefan at tribalmedia.se
Tue Dec 30 19:41:54 UTC 2008


I would say the right way is to 
 
1) create a sim-sim versioning scheme, so sims can refuse/negotiate comms on version numbering
 
and
 
2) Include an optional filtering in the grid service on version numbers - so that grids simply does not report neighbours with wrong sim-sim version scheme; which lets regions construct trusted and versioned maps transparently.
We did something like #2 on tribal net to accomodate for the fast up/fast down scheme that is an integral part of it - offline or timed out regions were filtered out separately from being removed from database, to let regions disappear for a couple of hours without losing their 'spot'.
Best regards,Stefan AnderssonTribal Media AB
> Date: Tue, 30 Dec 2008 16:33:29 +0000> From: jjustincc at googlemail.com> To: opensim-users at lists.berlios.de> Subject: Re: [Opensim-users] Consequences of vicinity> > Diva Canto wrote:> > Justin Clark-Casey wrote:> >> Is it worth updating the interface version number in OpenSim.Framework.Servers.VersionInfo to ensure that this is > >> enforced automatically for grids?> >>> >> > > Yes, to some extent, but not completely. This is about sim-sim comms, > > not sim-grid comms. Therefore if you havion n ue a running sim connected to > > osgrid who's owner is on vacation, that sim will continue to run -- and > > maybe cause problems to the neighbors -- even if you bump up the version > > number in Servers. We would need to check the version number on > > inter-region comms, not just on region registration with the grid. Or > > else on region registration, the grid server should look into the > > VersionNumber of the potential neighbors and warn/disallow registrations > > when the version numbers are not compatible.> > In principle, a grid server that has been upgraded could, on restart, kick off regions that have an interface number > that is now too low. This would be for a grid administrator that wants to ensure a smooth inter-region experience on > the grid.> > In practice, a grid administrator might want to be more lenient (perhaps notification rather than kick).> > I guess you could have a grid that allows different clusters of regions where those regions use a different/old > interface than other region clusters on the grid, yet all clusters used the same grid comms. The problem of teleport > (in which you still travel between disconnected clusters) might require a different interface version than an interface > for neighbouring sims.> > Nonetheless, in practice people who don't read these lists will continue to put sims side by side. If the version > number isn't bumped we may well see an upsurge in questions and confusion which we saw not long ago when OpenSim > revisions crossed interface incompatibility thresholds (whether grid <-> sim or sim <-> sim).> > Perhaps this is something that grid admins can give input on. Do you think the interface version number should be > bumped with neighbour interface changes to stop people having problems when they are neighbouring regions with > incompatible versions? Or should people always be instructed to keep regions away from each other unless they are sure > that their neighbours are going to update in sync with themselves.> > -- > justincc> Justin Clark-Casey> http://justincc.wordpress.com> _______________________________________________> Opensim-users mailing list> Opensim-users at lists.berlios.de> https://lists.berlios.de/mailman/listinfo/opensim-users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://opensimulator.org/pipermail/opensim-users/attachments/20081230/745b672a/attachment.html>


More information about the Opensim-users mailing list