[Update 2007-03-19@21:20] Right… good thing we only did 497 regions. We need to revert those regions back to 220.127.116.11992. We apologize for the inconvenience, but the affected regions will need to be restarted a second time to compelte the revert. It will be done in ca. 45 minutes.
[Update 2007-03-19@21:15] The deploy to 497 regions is done. The new version is 18.104.22.168788. You can find out which server version the region you are in is running by Looking at “About Second Life” from the “Help” menu.
[Update 2007-03-19@19:50] The blog post said that the Havok4 beta regions would be updated separately… and that separate update is now complete. The first roll of ~500 regions on the main release will begin in the next hour.
[Update 2007-03-19@12:45pm] Central system updates are complete. We took things slowly and also repaired some login servers that were misbehaving last night and had to be taken out of rotation. Anyone watching the online resident counts (“concurrency”) will see 3 brief dips between about 12:10 and 12:45 when the services that track online status were restarted. The numbers bounce back almost immediately as the system recovers, as designed.
We’ll be updating the server code on the hosts that make up the Second Life service this week to server code version 1.19.2 (current version is 1.19.1). As has been the case for quite some time now, this will be done via a rolling restart spread over several days, and there will be no extended downtime or new viewer required.
The list of changes is fairly small, and there should be no noticeable effect in-world:
- Support for PRIM_GLOW in llSetPrimitiveParams – this will require a forthcoming viewer update to author and compile scripts that use this, however
- Additional internal web service infrastructure work to remove dependencies on central databases (initially disabled, we’ll then slowly start flipping services on)
- Additional internal logging and diagnostics to track down the cause of various failures
- Fix for a rare “I just saw someone else’s IM!?!?!” bug
Our tentative schedule is as follows:
- Update affected central systems Tuesday morning [Update 2007-03-18@5:45pm] – this got started late while we tracked down a bug in the new code. This is halfway through, and will resume on Wednesday morning
- Update 3 racks (approximately 500 regions) on
Tuesday eveningWednesday evening Update 10 racks (approximately 1500 regions) on Wednesday evening Thursday morning
Restart remaining 15000 regions on Thursday eveningFurther updates are on hold until we resolve the problem discovered in rolling to the first 497 regions.
This will not affect the approximately 500 regions running in the Havok4 Beta program; they will be updated separately.
Region restarts will occur in machine order. Each region will be given a 5 minute warning, then restarted.