The following notes are taken from the Server Beta User Group (SBUG) meeting held on Thursday, January 28th, 2015, and the TPV Developer meeting held on Friday, January 30th. A video of the latter is included at the end of the article (my thanks as always to North for recording it and providing it for embedding), and any time stamps contained within the following text refer to both it and the TPV Developer meeting.
Server Deployments Week 5 – Recap
- On Tuesday, January 27th, the Main (SLS) channel received the server maintenance package previously deployed to Magnum and LeTigre
- On Wednesday, January 27th, the RC channels should all receive a new server maintenance package comprising an Experience Keys / Tools related fix, and internal Experience Keys / Tools improvements.
It was stated during the TPV Developer meeting that this update saw the support for the Avatar hover Height (AHH) capability extend to all three RC channels, although no mention of it was made in the release notes.
Tool Chain Project viewer
The Lab issued a new Tools Update project viewer, version 18.104.22.1688587, on Wednesday January 28th. This viewer has been built using the new viewer tool chain for Windows and Mac builds (which includes the upgrade to Visual Studio 2013 for Windows and, Xcode (5 or 6) for Mac, and the use of the autobuild 1.0 process).
Wiki notes for using the new build tools are in the process of being written / enhanced for self-compilers – please note these are a work in progress. Notes on the new autobuild process can also be found on the wiki.
[06:42] It is anticipated that this viewer will progress through to release candidate status, quite probably in week #6 (week commencing Monday, February 2nd).
Maintenance RC Viewer Issues
Users receiving the Maintentance RC, version 22.214.171.1248030, released on Tuesday, January 27th, have reported a series of irritating bugs with the viewer:
- BUG-8324 (and duplicate BUG-8333) reports the appearance of the term “nolink>” in numerous places within the viewer – such as toasts and notifications – this issue has been imported by the Lab and a fix is apparently in development
- BUG-8331 – some links containing the text secondlife or lindenlab incorrectly open a Place Profile in the Places panel when clicked
- BUG-8328 – very jagged pictures created by the snapshot floater
- BUG-1028 – the fix for MAINT-2056 “Lots of CURL GET FAILED spam in logs when browsing World Map – Beta viewer only” appears not to have fixed the problem, but rather replaced it with an equally spammy message.
Issues with this viewer can be tracked on the JIRA using this filter.
Viewer-Managed Marketplace (VMM)
[00:12] Brooke Linden was at the TPV Developer meeting to provide an update on the state of play with the Viewer-Managed Marketplace (VMM) work, which has been undergoing some beta testing with the project viewer on Aditi.
As a result of this testing, the Lab has been re-visiting the project viewer and incorporating feedback and observations for improvements into it, and are “really close” to having an updated version of the viewer available. When the appears depends on whether the Lab opt to release the updated viewer for limited beta testing on Aditi, or to wait until they’re ready to go with beta testing on the main grid, which is also not too far over the horizon.
The current plan is for more widespread beta testing to commence some time after February 14th (so as not to clash with any Valentines Day sales and promotions merchants may be running between now and Feb 14th), then run the beta for around a month, possibly longer, depending on feedback / issues. After this, VMM support will be deployed as a production tool, and things will move to a migration period which will also last around a month.
All this is still TBC, but a schedule will be posted (presumably via the Commerce / Merchant’s forum) once the Lab has worked through dates and requirements.
Experience Keys / Tools
[03:50] The Experience Keys / Tools project is currently awaiting some final server-side updates to be fully deployed (see the server release notes above). It is not anticipated that the Experience Tools viewer which is currently in the viewer release channel (version 126.96.36.1998001) will require any further significant updates, so its promotion to the de facto release viewer is pending the server-side updates.
In the meantime, the Lab has already commenced work on a set of enhancement to the Experience Keys / Tools which will further extend capabilities once they are ready to be deployed.
Avatar Hover Height Project
As noted above, server-side support for the Avatar Hover Height project (see my overview for details) is now on all three RC channels, and the plan is to deploy the server code to the Main channel in week #6 (week commencing Monday, February 2nd). Because of this, the current project viewer (currently version 188.8.131.528129) is liable to be superseded by a release candidate viewer following the server-side deployments for that week.
[18:00] A couple of small issues have been noted with AHH, and which only occur is very specific instances. One is that if an adjustment leaves your avatar floating more than around 1/2 a metre above the ground, an attempt to sit down can result in your avatar going up. In the other, standing on a prim and lowering your hover height into the prim can result in your avatar’s legs buckling as you slide into the prim. On terrain, where there can be limited physical penetration, this can be expected behaviour; but when occurring on a prim, it is suspected that the system may not be accurately assessing your avatar’s position.
[19:38] There have been one or two points of concern raised about the similarity in the name of this function – Avatar Hover Height – causing people to confuse it with the existing avatar hover slider (which will not be retired or removed from the viewer once AHH is fully deployed, as the two capabilities provide different use case solutions). The concerns are being fed back internally at the Lab, and if a name change is thought to be required, it may be slipped into the upcoming RC rebuild for the viewer, or the Lab may opt to wait and see if any confusion does arise.
There was a further group chat test carried out during the SBUG meeting on Thursday, January 29th, which revealed some issues of delays and severe viewer locking when attempting to switch from one conversation tab (e.g. local chat) and a group chat tab undergoing heavy use (and with large amounts of text being sent across it).
[06:00] The Lab has also been releasing further back-end updates as a part of their investigations into server-side lock-up and freezes, and they believe they are making good progress on tracking doing the underlying cause(s). However, given the frequency with which these freezes occur (it can be more than a week between servers locking-up), it may be some time before the Lab is more fully in a position to feel the problem can be / has been resolved.
SL Voice Issues
[12:22] The Lab has been working with Vivox to investigate the issue where a region has to be re-started after a rolling restart in order for voice to function. As a result the Lab has made some back-end changes which should ensure the voice channels restart properly when a region is restarted, but the results haven’t as yet been examined in detail to note any changes / improvements in the situation.
[12:48] There is a new set of SL voice files which are available for TPVs to pick-up and use; this has bugs fixes for reliability and quality and includes capabilities for producing voice quality metrics and connection error detection which should allow the Lab and Vivox to better diagnose issues when they occur. Unfortunately, there are still no Linux package forthcoming from Vivox.
[07:35] Izzy Linden, representing the Lab’s support team, frequently attends TPV Developer meetings to take input from the TPVs on matters of support and to provide feedback and information to them. However, a suggestion has been put forward to develop a means by which LL support staff and those supporting TPVs can communicate back and forth on major issues as they happen.
The Lab has in the past been resistant to such activities because they can lead to actions being taken outside of normal channels / processes, which can itself lead to problems. However, the idea here would be purely for matters of communication and passing information back and forth between support teams, rather than on seeking actual support.
The idea put forward would be for an invite only in-world group to be established where those noticing specific issues could post notice of the problem (e.g. receiving widespread reports from users on a voice issue or similar), so that all other support groups are alerted to the problem. Izzy has agreed to pass this back up the line to senior staff for consideration.
Inventory Fetching / Rezzing Issues
[11:37] The Lab has noticed an uptick in rezzing issues (failure to rez, etc.). The problem is being worked on, so there may be some upcoming maintenance to deal with it, although this has yet to be confirmed.
There have also been reports of inventory load failures by Firestorm users, although nothing has (at the time of the TPV Developer meeting) been similarly reported by users on the Lab’s viewer. In fact, the Lab have been seeing something of a reduction in the number of reported inventory load failures.
SL Go SLV Teleport Issue
As I reported on Friday, January 30th, those using the SL Viewer (SLV) within the the SL Go service (on PC, Mac, Android or iPad), are currently finding they are unable to teleport. While it is unclear why this happening or where the problem lay, OnLive have decided to present all Sl Go users with a free month’s subscription for any inconvenience they may have suffered.
[19:50] As a part of on-going investigations, the Lab’s QA team have turned up the logging and have been testing the service themselves in order to generate stats on the failures. They believe they have found the point where things go “really wrong”, which lies within the handshaking between the two simulators involved in the teleport attempt.
However, what remains unknown is precisely what is going wrong between the SL Go SLV, and the simulators in order for the simulator / simulator messaging to go so wrong. OnLive’s engineers are also trying to isolate any possible cause for the issue (which does not affect users running Firestorm on SL Go) as well.