Justin Clark-Casey wrote: > On 21/10/10 23:57, John Hopkin wrote: > > Thanks, Justin. Do you happen to know if the first save will be > > reliable (other than the missing assets)? It's that word "aborted" > > that worries me. > > You're right to be worried - the first OAR will not be usable. I have added > some text to the error message to make that > clear.
Thanks. > > And how can I go about removing references to the missing assets? Is > > this a case of hacking the MySQL database? > > No, unfortunately it's more complicated than that. The missing assets could > be referenced anywhere (within serialized > objects, asset metadata or even hardcoded within scripts). > > In principle, you could untar the successfully saved OAR and grep all the > files for the missing asset UUIDs. That would > highlight where they are being referenced. > > But really, the missing asset problem should not be causing your initial OAR > abort. The asset service should return > notification of missing assets as quickly as it returns actual assets. Are > you using a stock asset service? Also, how > many assets actually end up being saved - is it a very populated region? The asset server is OSGrid's. One of the regions affected has around 8100 assets, the other about 230. I think there's another one, but I'm not sure which of the 13 I have it is. > > Justin Clark-Casey wrote: > > > >> OAR save times out after 60 seconds if it doesn't receive all the assets > >> it's asked for. This is to prevent completely > >> or very unresponsive servers from keeping the file open indefinitely. > >> > >> The errors you're seeing suggest that the assets not being delivered are > >> missing anyway, otherwise they would get into > >> the subsequent save. > >> > >> On 21/10/10 00:35, John Hopkin wrote: > >>> Finding this a lot while doing "save OAR" (text in<> is variable): > >>> > >>> 00:10:27 - [ARCHIVER]: Asset service failed to return information > >>> about<n> requested assets > >>> 00:10:32 - [ARCHIVER]: No information about asset<UUID> received > >>> [repeated<n> times] > >>> 00:10:32 - [ARCHIVER]: OAR save aborted. > >>> [...] > >>> 00:10:53 - [ARCHIVER]: Received information about asset<UUID> after > >>> archive save abortion. Ignoring. > >>> [repeated<n> times] > >>> > >>> So, I'll get several "No information" lines, followed by "OAR save > >>> aborted" and, a few seconds later, several "received information" > >>> messages. > >>> > >>> The OAR checks as complete with gzip, but I don't know if it's > >>> complete or not. When performing several "save OAR"s in a command > >>> script, it's not clear which region is affected, or even that there > >>> has been an error at all unless one looks for it. > >>> > >>> This could lead to incomplete saves being archived. Obviously, this is > >>> dangerous. > >>> > >>> On rerunning the "save oars" during the same OpenSim.exe session, I > >>> get the message: > >>> > >>> [00:29:35] - [ARCHIVER]: Could not find asset<UUID> > >>> > >>> for each of the above lines . This time, there is no "OAR save > >>> aborted" message; presumably, the OAR worked. > >>> > >>> Any ideas? > >>> > >>> OpenSim v0.7.1 8731c2b, OS is Ubuntu Meerkat latest; database is MySQL > >>> 14.14; Mono is 2.6.7; physics is Meshmerizer/ODE. -- John Hopkin _______________________________________________ Opensim-users mailing list [email protected] https://lists.berlios.de/mailman/listinfo/opensim-users
