Collins Land, September 2013 (Flickr) – blog post
The following notes are drawn from the Server Beta user group meeting held on Thursday, November 13th, the transcript for which can be found here.
Server Deployments Week 46 – Recap
As always, please refer to the server deployment thread in the forums for the latest information and updates.
- There was no deployment to the Main (SLS) channel on Tuesday, November 11th.
- On Wednesday, November 12th, all three RC channels received the same server maintenance package, which comprises “minor improvements” to help configure the texture and mesh CDN, by allowing the Lab to reconfigure the CDN URL if they need to, with the intention of the of making it a more dynamic host name in the future.
Snowstorm RC Viewer
A new Snowstorm contributions RC viewer arrived on Thursday, November 13th. Version 18.104.22.1686724 promotes the viewer from project status, and brings with it the following contributed updates:
- OPEN-215 kCGLRPTextureMemory is a deprecated function as of Mac OS X 10.7, replace with kCGLRPTextureMemoryMegabytes
- OPEN-268 FFLOAD_XML missing on linux and darwin, FFSAVE_XML missing on darwin
- STORM-2053 applicationShouldTerminate function returns NSApplicationDeligateReply when it should return NSApplicationTerminateReply
- STORM-2056 Projector reflections do not respect the environment intensity parameter
- STORM-2067 Glossy Projectors
- STORM-2069 Delete key doesn’t delete first character of marked text on OSX
- STORM-2070 Sticky modifier keys after OS window comes up on OSX
- STORM-2071 Unwanted InputWindow comes up when typing Japanese on OSX
- STORM-2072 Bad behavior of Input Window
- STORM-2078 Editing an objects rotation with the rotation rings often causes the object to jump to position <0,0,0> on the region and rotation changes to <0,0,0>
- STORM-2080 Dresses purporting to be Fitted Mesh stretch to 0,0,0
- STORM-2081 Second Life 3.7.18 (295539) Oct 16 2014 08:19:54 (Second Life Release) crashes every time upon viewer Quit under OS X Yosemite 10.10
Attachment Fix and Maintenance RC Updates
The current RC viewers – the Attachment fix RC viewer and the Maintenance RC viewer both updated on Wednesday, November 12th as a result of the Benchmark viewer promotion to release status. The updates were as follows:
- Attachment Fix RC viewer updated to version 22.214.171.1246729. This viewer adds some fixes to previously released changes in the way joint offsets in rigged meshes are handled & fixes some issues found with adding and removing attachments after the recent AISv3 deploy, and improves the status information shown in inventory for attached objects
- Maintenance RC viewer updated to version 126.96.36.1996734. This viewer offers a broad range of fixes for voice, privacy, rendering, texture animation, avatar distortion, inventory management, sounds, Mouselook in Mac, multiple UI fixes in script editor, Pay flow, chat, stats floater, edit menu, etc.
Release Viewer VFS Failure Issue
It’s still not clear how widespread the VFS failure issue on the current release viewer extends (see part 1 of this week’s update, and BUG-7776). As noted in the first part of this report, I’ve managed to resolve such problems by manually deleted the viewer cache, and this worked for me in this case. Whirly Fizzle posted the same advice to the bug report, but at least one person has indicated it didn’t resolve the issue for them.
“Fast Pipe” Viewer
Monty Linden is working on further viewer-side updates to address reported problems being experienced in region rezzing times (such as reported in BUG-7698). There’s an initial release of the viewer available to those who can self-compile, and Whirly Fizzle reports that the fixes appear to work, commenting that, “Fast-pipe with pipelining enabled def works better for me on CDN regions. The long texture loading stalls have pretty much stopped.” There’s no date as to when this viewer will publicly appear as an RC or project viewer.
Viewer Stats Ping Sim Data
There have been a few question in both meetings and in the forums about what the Ping Sim data in the viewer’s statistics bar actually represents. According to the wiki page:
Ping Sim: How long it takes data to go from your computer to the region you’re currently in. This is largely dependent on your connection to the Internet. If Ping Sim is high but Ping User is not, the server might be having problems.
However, this doesn’t paint a complete picture, as there can be a noticeable difference in the ping sim value given in the statistics bar of the viewer compared to pinging a simulator host directly. Responding to a question during the Server Beta meeting, Maestro Linden said:
As I recall, the ping number is your actual network round trip, measured by the UDP connection to the sim, *plus* some other time, like the time required to render the frame… So if you’re only getting 10FPS, that’s an automatic +100ms on top of the actual network round trip time… That ping time may also include the sim’s frame time (22.2ms if running at 45fps).
Answering a similar question through the forums in October, Oz Linden offered further clarification:
Looking at the code (I had not had occasion to look at this before), the Ping Sim measure appears to be based on a separate Ping message (our own message type transmitted over UDP, not ICMP). Those messages are mixed in periodically with the other UDP messages that are more or less constantly flowing between the viewer and the simulator. Because it’s the application that is turning that message around rather than a low-level part of the network stack, the fact that it is consistently higher than ICMP ping to the same host isn’t surprising.
[ICMP, or Internet Control Message Protocol is the protocol generally used by ping operations.]
A couple of upshots of this is that if there is packet loss as a result of packet loss with increase ping rates as seen in the viewer, while no actual increase is occurring in the network connection itself, while increasing traffic within in region will cause higher ping rates within the viewer, simply because of the resultant UDP packet queuing in the server (hence a possible reason for BUG-7797, where the ping increase is seen as causing an increase in “lag”, rather than being indicative of an increase traffic load occurring). As such, Maestro commented that the viewer’s ping data is, “probably a more useful measure of latency than network ping.”