[Opensim-users] Intermittent problem in Serverless Grid mode

Ovi Chris Rouly maelzel at ieee.org
Sun Jun 6 20:11:24 UTC 2010


Hello everyone,

I'm using OS 0.6.9 compiled from source and running on Win7 64-bit.  Currently, I am operating in HyperGrid-StandAlone (Serverless Grid) mode.  Although, over the last two years, I've run OS in each of its supported modes at various times, respectively.

When I start my current simulation, I launch a 4X4 matrix of 16 OpenSim.32BitLaunch.exe instances.  The OS Simulator instances are themselves each a 4X4 matrix of 16 Regions.  Thus, total adjacent (contiguous) regions are 256.  I typically use the SL viewer to visualize my simulation.  However, the problem I'm about to describe also existed on XP SP2 64-bit and does exist with the Hippo viewer.  Please note, I run a LAN connected between the "server-side" box and 2 separate "client-side" boxes.

Once my simulation is launched, I can generate my 16, 32-bit OS Sims and I can see within the SL map viewer all of the 256 Regions without any trouble.  Aside from the problem I'm about to describe, I can move my avatars and their inventory (custom JPG tattoo-skins and tiny body-sizes) through-out the entire simulation.  No data is ever lost through the MySQL backend and simulations with OS 0.6.9 are actually quite good otherwise.

I have 2 intermittent OS problems.  I will describe only one here.

Sometimes, the mini-map output and the simulated "physical" terrain output associated with the mini-map display does not exist after a Teleport to a distant region.  Effectively, after TP, my avatar will reappear suspended over water and then a few seconds later, it will just disappear.  I may, or may not, see adjacent Regions within the SL viewer.  My Regions are numbered 0-15 on a side; 0-3, 4-7, 8-11, 12-15, per OS Sim instance, respectively.  Ports, of course, are unique.

This is an intermittent failure and can (sometimes-by-work-around) be overcome by Teleporting or Flying/Walking into an adjacent-diagonal Region and then re-entering the missing area "straight-on."  (Note: it appears that Regions only generate when I'm +2 Regions removed from the offending "target" Region.)  However, when this problem does occur, even this "work-around" will not cause the missing Regions to generate.  Furthermore, there is no consistency as to which regions will fail to rezz and demonstrate this problem.  Even my "home" region can demonstrate this trouble when I TP back to it.

Time to generate-rezz is not an issue.  I have literally waited overnight in tests to understand this problem.  Also, if the missing Regions "ever do" generate, they can be manipulated and WILL hold prims for later usage.

Has anyone seen anything like this?  If it has already been reported in a Mantis, I couldn't find it.  Thanks in advance for the opportunity to ask the user group.  I look forward to your collective help.

Respectfully,

Chris
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://opensimulator.org/pipermail/opensim-users/attachments/20100606/f0aa42d8/attachment.html>


More information about the Opensim-users mailing list