Mantis Bug Tracker

View Issue Details Jump to Notes ] Issue History ] Print ]
IDProjectCategoryView StatusDate SubmittedLast Update
0007741opensim[REGION] Physics Enginespublic2015-11-17 14:302015-11-28 08:47
ReporterGarmin Kawaguichi 
Assigned To 
PriorityhighSeveritycrashReproducibilityalways
StatusnewResolutionopen 
PlatformOSOS Version
Product Versionmaster (dev code) 
Target VersionFixed in Version 
Summary0007741: ODE: starting with existing regions generates a fatal error
DescriptionWith Opensim and ODE (the old:
    meshing = Meshmerizer
    ; meshing = ubODEMeshmerizer
    physics = OpenDynamicsEngine
    ; physics = ubODE)
the console displays errors as in attachment 1. At the same time, Windows creates as many error windows as there are regions (attachment 2). The text of the error window is very particular: "assertion "false" failed in
c:\avination\libode\ode-131\src\threading_fake_sync.h:83" As such a path does not exist anywhere on our LAN, it is obvious that it is hard coded in Opensim.
It is not found in the Opensim source code, but in ode.dll in both lib32 and lib34. In lib32 at 0007:87db
Note: in Opensim.log no trace of error; the file ends with:
2015-11-17 18:12:50,989 DEBUG - OpenSim.Region.CoreModules.ServiceConnectorsOut.Grid.RegionCache [REGION CACHE]: (on region VarVide1) Region Vide8 is up @ 5002-5001
2015-11-17 18:12:50,989 DEBUG - OpenSim.Region.Framework.Scenes.SceneCommunicationService [SCENE COMMUNICATION SERVICE] Region Vide8 successfully informed neighbour VarVide1 at 5003-5000 that it is up
2015-11-17 18:12:50,989 DEBUG - OpenSim.Region.CoreModules.ServiceConnectorsOut.Grid.RegionCache [REGION CACHE]: (on region Vide8) Region VarVide1 is up @ 5003-5000

MML
TagsNo tags attached.
Git Revision or version numberopensim-0.6.9.rc1-16006-geab427e
Run Mode Standalone (Multiple Regions)
Physics EngineODE
Script Engine
Environment.NET / Windows32
Mono VersionNone
Viewernot needed
Attached Filestxt file icon Attachment 1 the errors in the console.txt [^] (3,180 bytes) 2015-11-17 14:30 [Show Content]
png file icon Attachment 2 20151117_opensim-0.6.9.rc1-16006-geab427e_ODE_error.png [^] (69,239 bytes) 2015-11-17 14:31


zip file icon Attachment 3 OpenSim.log.zip [^] (69,069 bytes) 2015-11-18 08:08

- Relationships

-  Notes
(0029551)
UbitUmarov (administrator)
2015-11-17 15:19

c:\avination\libode\ode-131\src\threading_fake_sync.h:83 is a debug path stored at compile time on the ode.dll library.
The cause of that problem is a lot hard to identify from the information you provide :(
(0029552)
UbitUmarov (administrator)
2015-11-17 15:24

what does the opensim-0.6.9.rc1-16006-geab427e mean??
what is really the opensim version you are using?
(0029553)
abitar (reporter)
2015-11-17 15:26

One thing you can try is make sure it recompiling the script binaries. I find if there to much done with the scripting engine you have to delete the temp files for the scripting engine since the format is not right for it to be loaded. But that would means you loose the states of all the current scripts.

Also, do you get an error in world, it might point you to the faulty script.
(0029554)
Garmin Kawaguichi (reporter)
2015-11-17 15:58

opensim-0.6.9.rc1-16006-geab427e = the name of the Opensim master in the shortlog (2015-11-17 09:00 UbitUmarov work around some 'tests' errors: UUID.Zero is a i )
and
the zip container is opensim-0.6.9.rc1-16006-geab427e.zip

We noted the curious version number (0.6.9.rc1), but we assumed that it was still an Avination release number; we thought you change it when you have time.
----------------------------
No viewer connected at that time; it is Opensim startup. Windows removes the thread after few seconds.

I only compiled Opensim from the code by changing VersionInfo.cs flavour
For that test, regions are 9 standard and 4 vars.

MML
(0029555)
UbitUmarov (administrator)
2015-11-17 16:12

Ok sounds like something on the web server that manages the binaries
Plz provide me the log prior to the errors, so i see the context.
Region stopped? or just kept firing those errors?
what is yr windows version?
(0029556)
aiaustin (developer)
2015-11-18 06:38
edited on: 2015-11-18 08:34

The path/file naming issue seems to come from a problem with the .tar and .zip download files on the main OpenSim "ViewGit" entries at
http://opensimulator.org/viewgit/?a=shortlog&p=opensim [^]

When you click on the tar or zip icon it now gives this file name to save... for example for the latest dev master....

opensim-0.6.9.rc1-16010-g0fbb4f0.tar
opensim-0.6.9.rc1-16010-g0fbb4f0.zip

(0029558)
Garmin Kawaguichi (reporter)
2015-11-18 08:05
edited on: 2015-11-18 08:07

Attachment 3 is the Opensim.log file matching errors defined by attachments 1 & 2.

MML

(0029559)
Garmin Kawaguichi (reporter)
2015-11-18 08:32
edited on: 2015-11-18 08:33

About Windows:
Windows 10 Professionnel 32-bit

Immediately after the error (and I had not answered the error windows), Windows removed the application Opensim.

Next test: same bin, same config, same regions but with scripts removed.

MML

(0029560)
Garmin Kawaguichi (reporter)
2015-11-18 12:32

A) Tested without script(s); the regions have been cleaned until there are no more scripts (change region root and show scripts reporting no scripts): the errors are the same described in the mantis description.
B) With meshing = ubODEMeshmerizer and physics = ubODE, the migration from an old ODE (with scripts) to an uBode physics presents no issues (at this stage of the migration test).
C) With meshing = Meshmerizer and physics = BulletSim, the migration from an old ODE (with scripts) to a BulletSim physics presents no issues (at this stage of the migration test).

1) "(at this stage of the migration test)" These tests are very simple, just an attempt to migrate a set of lightly loaded regions; when we'll try the multiple scenarios that simulate our real OpenSimulator applications, I could have a more detailed opinion to give you. For now, I'll wait until you improve migration.
2) By "an old ODE" I mean simulators were working properly with OpenSimulator 0.8.2 (or almost :)

MML
(0029561)
UbitUmarov (administrator)
2015-11-18 12:46

My guess at the moment is that it may be related to several regions on same instance and on both ode engines ( ode.dll issue )
(0029562)
UbitUmarov (administrator)
2015-11-18 12:52

One test you can do.. setting the conditions that cause the problem...
replace the bin/lib32 by a ode.dll from master before the merge..
( or lib64 if 64bits )
(0029569)
Garmin Kawaguichi (reporter)
2015-11-19 07:47

1) Test with an old ode.dll (32 bits) copied from an old master (2014) no problem

2) Additional tests (with the new ode.dll of course):
Regions *.ini contain:

std1 no problem
std1 std2 no problem
var1 no problem
var1 var2 no problem
std1 var1 errors
var1 std1 errors

MML
(0029576)
UbitUmarov (administrator)
2015-11-19 16:59

I pushed a change on ubOde, can you please test it with the crash conditions? / new ode.dll etc)
( old ode not changed still)
(0029583)
Garmin Kawaguichi (reporter)
2015-11-20 05:19

This Mantis was for ODE; with ubOde and Bulletsim there was no errors!
(0029585)
UbitUmarov (administrator)
2015-11-20 07:02

Ooops
(0029587)
Garmin Kawaguichi (reporter)
2015-11-20 08:59

Tested with opensim_20151120_1548_ad6080b same protocol: always errors as above.

MML
(0029727)
UbitUmarov (administrator)
2015-11-27 14:53

added some code that may fix that issue
(0029748)
Garmin Kawaguichi (reporter)
2015-11-28 08:47

@UbitUmarov: test with 118 g9928076 from 20151127 2346

- the ODE Internal Error 1 avination etc: solved

- opensim ODE error: no

- opensim ubODE error: no

During the first run I had an error with meshmerizer
Opensim.Region.PhysicsModule.Meshing.Meshmerizer.Meshmerizer.cs Ln 1011 CreateMesh(...) dupplicate key; unreproducible.

MML

- Issue History
Date Modified Username Field Change
2015-11-17 14:30 Garmin Kawaguichi New Issue
2015-11-17 14:30 Garmin Kawaguichi File Added: Attachment 1 the errors in the console.txt
2015-11-17 14:31 Garmin Kawaguichi File Added: Attachment 2 20151117_opensim-0.6.9.rc1-16006-geab427e_ODE_error.png
2015-11-17 14:38 Garmin Kawaguichi Description Updated View Revisions
2015-11-17 14:39 Garmin Kawaguichi Description Updated View Revisions
2015-11-17 15:19 UbitUmarov Note Added: 0029551
2015-11-17 15:24 UbitUmarov Note Added: 0029552
2015-11-17 15:26 abitar Note Added: 0029553
2015-11-17 15:58 Garmin Kawaguichi Note Added: 0029554
2015-11-17 16:12 UbitUmarov Note Added: 0029555
2015-11-18 06:38 aiaustin Note Added: 0029556
2015-11-18 06:42 aiaustin Note Edited: 0029556 View Revisions
2015-11-18 08:05 Garmin Kawaguichi Note Added: 0029558
2015-11-18 08:07 Garmin Kawaguichi Note Edited: 0029558 View Revisions
2015-11-18 08:08 Garmin Kawaguichi File Added: Attachment 3 OpenSim.log.zip
2015-11-18 08:21 aiaustin Note Edited: 0029556 View Revisions
2015-11-18 08:32 Garmin Kawaguichi Note Added: 0029559
2015-11-18 08:33 Garmin Kawaguichi Note Edited: 0029559 View Revisions
2015-11-18 08:33 aiaustin Note Edited: 0029556 View Revisions
2015-11-18 08:34 aiaustin Note Edited: 0029556 View Revisions
2015-11-18 12:32 Garmin Kawaguichi Note Added: 0029560
2015-11-18 12:46 UbitUmarov Note Added: 0029561
2015-11-18 12:52 UbitUmarov Note Added: 0029562
2015-11-19 07:47 Garmin Kawaguichi Note Added: 0029569
2015-11-19 16:59 UbitUmarov Note Added: 0029576
2015-11-20 05:19 Garmin Kawaguichi Note Added: 0029583
2015-11-20 07:02 UbitUmarov Note Added: 0029585
2015-11-20 08:59 Garmin Kawaguichi Note Added: 0029587
2015-11-27 14:53 UbitUmarov Note Added: 0029727
2015-11-28 08:47 Garmin Kawaguichi Note Added: 0029748


Copyright © 2000 - 2012 MantisBT Group
Powered by Mantis Bugtracker