There was lots of detailed information in the third-party dev UG meeting last week. We got a bunch of information on what is happening with a couple of problems. Currently, disconnects and detachments are front lines.
Servers
The main channel rolls to #19.03.15.525315. This is the OS upgrade. It includes the EEP support.
The RC channels Blue Steel, Le Tigre, and Magnum will be the first version with UDP asset fetching disabled or removed. See the Tools & Tech section of the SL blog Obsolete Asset Fetch Disabled for more details.
The UDP change will mean older viewers using UDP asset fetching will stop working correctly. Avatars are likely to not render or render as Ruth/Roth. Sooner or later you will have to update…
Disconnects
The RC channels will have a change that is hoped will help with the disconnect problems people are experiencing, which seems to have started in December or January. Some think EEP is causing the disconnects. EEP is coincidental with the disconnects. The current analysis of the problem suggests it is in the TCP communication between the viewers and servers.
In the March 29, Third-Party Dev UG meeting Oz talks about the problem and why it took the Lindens so long to notice. Video Basically, the problem is not showing in the performance stats the Lindens watch.
The disconnects seem to fall into groups, those occurring during a teleport and those that are periodic (disconnect every so many minutes). People trying to chase the problem down are having problems reproducing the disconnect. They have the problem rated as #1 & #2 and have lots of people on it full time plus a QA team of remote users, meaning far from the Lab distance-wise.
There are rumors about different things one can do to stop the disconnects, change the WL/EEP sky, cancel a TP mid-teleport (cancel, open World Map – find destination region – wait for it to come up and you pop in), and others. For others the key is location, Blake Sea is a problem and Heterocera is not a problem. Continue reading