Deploy Plans for the week of 2019-10-21

[Caleb Linden]

Note: https://releasenotes.secondlife.com/simulator/2019-10-03T01%3A12%3A11.531528.html goes everywhere

 

Second Life Server:
https://releasenotes.secondlife.com/simulator/2019-10-03T01%3A12%3A11.531528.html

Second Life RCs:
https://releasenotes.secondlife.com/simulator/2019-10-03T01%3A12%3A11.531528.html

Scheduled Wednesday 2019-10-23 07:00-10:30 PST

 

On Region Restarts:

Regions will be restarted if they have been running for more than 10 days on Tuesday (Main Channel) and Wednesday (RCs) regardless of whether or not new code is being deployed, for the general health and well being of the Simulators. Nothing beats turning it off and then on again ... once in a while.

As always please plan events around these maintenance windows. However, if you are hosting an event on Tuesday or Wednesday when we're not deploying new code, you can restart your region ahead of time to avoid interruptions.

[Lucia Nightfire]

I look forward to testing the crash vector that most likely wasn't fixed.

[Miller Thor]

Why do we get rolled back to 531528 on Magnum-RC?

531693 works much better especially with the Internal Script Improvements. 

Edited October 23, 2019 by Miller Thor

[Qie Niangao]

8 hours ago, Miller Thor said:

531693 works much better especially with the Internal Script Improvements. 

Apparently not. Some very strange script-related bugs have been reported on 531693 and other releases newer than 531528. A flock of seemingly distinct jiras have been closed as duplicate to the one left standing. 

[Miller Thor]

Interesting. In long-term observations, we noted a positive stabilization and, consequently, an increase in script performance, which had more than a positive effect on the regional performance. From us, the last week rolled version could remain. And my wife and I know our regions only too well and know the performance to judge it.

With the todays DOWNGRADE we lost more than 50% of the Scriptperformance we had before with 531693. Is that the will of Linden to keep us on low Level?

Edited October 23, 2019 by Miller Thor

[Qie Niangao]

2 hours ago, Miller Thor said:

With the todays DOWNGRADE we lost more than 50% of the Scriptperformance we had before with 531693.

That's way too much. I've been watching script performance for a number of Mainland regions the past few months and have seen only single-digit increases in Scripts Run percentages with the recent script improvements. That may be all we should expect at this point because they've only really affected listen() and sensor() events -- there's been no attempt yet to address the elephant in the room: inefficient impact of idle scripts on scheduler performance.

That said, though, I too have seen a region's script performance dip by 50% with one restart and restore to the previous high level with the next even when all three conditions were on the very same release. This happens at random, as far as I can tell. If you can arrange a restart for your region it may be a magic cure -- until the next unlucky restart, and there's no telling how many attempts it takes to get a good host -- but it may be worth a try.