Over the next several days we’ll be updating the servers comprising Second Life to our “1.19.1” version of the server code.
These changes include a small number of bug fixes including a fix for VWR-882: Group name showing as (???) and some crash fixes, but most of the changes involve better stats logging or infrastructure for upcoming viewer work.
Here’s our tentative schedule for the update:
- Tuesday PM: Update “central” system – [COMPLETE]
WednesdayFriday AM: Rolling restart of ~500 regions – [COMPLETE] ThursdayMonday AM: Rolling restart of ~1500 regions ThursdayMonday PM: Rolling restart of the rest of the grid
Regions running the Havok4 Server Beta code will not be updated as part of this process.
Apart from the usual restart (after a 5 minute warning) that each region will experience, we do not anticipate other widespread effects on the grid, but we’ll be watching things closely and are prepared to revert the code changes in case of unforeseen behavior.
[Update: 4:47pm] Since it was asked in the comments: – no viewer updates are required as part of this work.
[Update: 6:00pm] Central services updated have been completed. There were some temporary and expected login failures as systems updated themselves, but our monitoring indicates everything went smoothly. Among the systems updated are those responsible for reporting concurrency, so the dip from 5:20pm PST to 5:45pm PST that would show up on concurrency graphs put together by residents (this means you, Tateru Nino!) doesn’t represent an actual drop in the number of residents online.
[Update: 2008-02-20 06:30am] Server version 126.96.36.199264 has been rolled to 521 regions.
[Update : 2008-02-20 08:50pm] We discovered some problems with the 188.8.131.52264 server release. The regions which received the updated code have been patched and are now operating properly; for more information, please see this blog post. However, we are putting the rest of the rolling restart on hold until we can better diagnose and fix the problem. We will post more information about the 1.19.1 server update as we have it.
[Update: 2008-02-21 2:54pm] The problem identified during the first phase of the 1.19.1 Server deploy turned out to be isolated and easily fixed (it only affected roll-backs to auto-saves). We’ll be resuming the deploy sequence, just shifted a few days later.
[Update: 2008-02-22 5:55am] Server version 184.108.40.206455 has been rolled to 512 regions.