From the server deployment thread for the week:
We are working on the TP & sim crossing disconnect issue, and making several changes over this week. These may be a little bit more disruptive than our usual grid roll process and we apologise in advance for the inconvenience of those. We’ll do our best to keep this disruption to a minimum. We have a simulator update which we will roll to BlueSteel and LeTigre on Tuesday. Depending on the results we will make additional plans for gridwide rolls as soon as practical. We thank you for your patience and fully realize both the urgency and the frustration this has been causing.
Thus far, there has been a deployment to BlueSteel and LeTigre – server update 19.04.22.526534. Check the deployment thread for further updates.
There have been no viewer updates at the start of the week, leaving the viewer pipelines as follows:
- Current Release version 220.127.116.116190, formerly the Estate Access Management RC viewer, dated April 12, promoted April 17 NEW. – see my EAM overview for more information
- Release channel cohorts:
- Project viewers:
- Linux Spur viewer, version 18.104.22.1689906, dated November 17, 2017 and promoted to release status 29 November – offered pending a Linux version of the Alex Ivy viewer code.
- Obsolete platform viewer, version 22.214.171.1240847, May 8, 2015 – provided for users on Windows XP and OS X versions below 10.7. This viewer will remain available for as long as reasonable, but will not be updated with new features or bug fixes.
Let’s see … for server news, many of us are focused on the teleport issues … I hate to promise anything, but we all know a lot more about TP internal problems than we used to 🙂 . We do understand they’re annoying and a frustrating problem – we definitely want to fix it and make it better.
– Simon Linden, Simulator User Group, April 23rd
From the above statement, it should be clear that the Lab is devoting a lot of time to teleport disconnects. There have been mixed reports on the outcome of the deployment made on Thursday, April 18th, 2019, with some indicating matters have improved, others reporting no real change. In particular, people visiting Fantasy Faire and travelling around the new Linden Homes continent appear to be faring a lot better than had been the case. Which is not to say the issue has in anyway been resolved – hence the continuing work.
While it had been indicated that the recent operating system update may have played a role in the problems, Simon again referenced the timing issue with region crossings, whilst also mentioning the potential for the simulator side of EEP possibly also playing a role in things.
One of the puzzles we’re trying to sort out is if somehow the environmental work caused the TP problems – the timing of the release is suspicious but the functionality _should_ be different …. The disconnects are a problem where the viewer and 2nd region don’t start talking as they should [so something is out of sync] or there’s a failure to communicate. I know, for example, the 2nd region is waiting for the viewer to connect and get a message … that never happens …
It doesn’t seem to be closely associated with AV complexity … that said, the more complex your AV is, the more work it needs to change regions. It’s always been better to have fewer scripts and data for teleports and region crossings.
The problem is still trying to pin down actual potential causes, with disconnects remaining inconsistent in terms of reproduction., again as Simon noted:
It’s frustratingly inconsistent. It’s a lot easier to fix something that breaks all the time … to fix it, and to know when you fixed it. For example, before this meeting we ran a test that had probably 400 or so successful teleports, no disconnects … that’s good, but not proof of a fix.
Mazidox Linden, from the Lab’s QA team further qualified the Lab’s problem:
We need in the neighbourhood of 10000 teleports to have any kind of real statistical confidence, just as a reference point 🙂 .
EEP / Windlight Issues
Following the Thursday deployment, many region holders / designers noted significant differences in how their region windlights were being rendered. The didn’t get much in the way of discussion during the meeting, however, Rider Linden offered an apology for the situation, noting that the focus on the TP issue(s) resulted in some unexpected regressions. However, it’s not currently clear what might be done to deal with this issue.