I learnt from another source today that I made a mistake in commit 690f30c and wasn't actually logging the exception.
I've corrected this in git master 432f0e8 so please update to this.
On 29/11/14 02:21, Butch Arnold wrote:
Hi Justin,
Thanks for the reply, sorry for the error on the links.
The non working pic is at: http://3rdrockgrid.com/new/testpics/WelcomeNW.jpg
I will check out the new code.
Thanks again.
On 11/28/2014 9:14 PM, Justin Clark-Casey wrote:
Hi Butch. You actually linked to the 'ok' pic twice below. It's good to see
the actual output as that can sometimes
give extra clues to someone looking at the code as to what's going wrong.
However, on reviewing the code, I believe it is possible for an exception in
the script compilation process to
permanently stop all future compilation until the simulator is restarted. I
also believe any such exception would be
silently dropped and not logged.
So to help identify the issue, I have both logged any exception and changed the
code to increase the chance that
future script compilation will still happen even if something does go wrong.
This is in commit 690fe0c in master.
If you see an exception logged with this code then please report. If the
problem still occurs with no extra log clues,
then my next hypothesis would be that compilation is hanging indefinitely,
possibly due to some Mono bug. In this
case, the ideal thing to do would be to take a thread dump so we can see if a
thread is stuck. On Mono I have just
documented the process at [1]. However, I suspect you're on Windows and I'm not
sure how one would do it there (or
even if that kind of thing is possible).
If we find the bug in master a fix can be backported to 0.8-post-fixes branch.
[1] http://opensimulator.org/wiki/Debugging#Thread_Dumps
On 28/11/14 16:46, Butch Arnold wrote:
Hello Justin,
Thank you.
A little more info...
It seems not "All" scripts stop working... those which were running will remain
running, but any that should start
"After" scripts break will not start - instead they are shown in the "xengine
status" report from the console as
"Scripts Waiting For Load" with the number of scripts waiting to load... see
screen captures here:
Working fine:
This is after a fresh restart of the instance....
http://3rdrockgrid.com/new/testpics/WelcomeWorking.jpg
This is immediately after we were able to break the same "Fresh Restart" from
above....
Broken:
http://3rdrockgrid.com/new/testpics/WelcomeWorking.jpg
This seems to happen on any of our instances as it should since they are all
configured the same.
When I upgraded, I did not use anything from the previous grid setup... ie.. I did not
"Merge files" - I built a new
grid beside the old one in case I needed to roll back to the old.
Everything is fresh in this setup.
I've installed a new database, but reloaded with a dump from the old one and
new folders with no previously compiled
scripts.
Our setup recompiles scripts on every instance start.
I thought maybe one of our custom .dll's were to blame so I removed them, running only
"Bone Stock" OpenSim and still
have the same issue.
- I have made the changes you've suggested with the same results.
- I have upgraded to the recent Ghost/Trunk merge 2 days ago with the same
results.
- I have tried "Stopping and Restarting" scripts from the console when the
issue starts... same results.
- I have tried "Suspending and Resuming" scripts from the console when the
issue starts... same results.
- The only way to get scripts running correctly again is to restart the
OpenSim instance.
It seems a 32bit OpenSim instance will break instantly... ie we have found a
combination which reliably crashes the
instance each time in 32bit mode.
If running a 32bit instance, any time a HyperGrid visitor arrives it causes the
issue.
In 64bit mode, we have not yet identified an event, a chain of events which
cause the issue.
Last evening, I made some changes and restarted the region in 64bit mode, we
tried several times to break scripts and
was unable to, so we were hopeful things were fixed.
We left the region run overnight and this morning I checked and scripts were
broken again.
Logs do not indicate any issues.
We are still trying to find the combination which will break scripts every time
in 64bit mode.
You can see it in action for yourself at: (providing it isn't already broken)
grid.3rdrockgrid.com:8002:welcome
We have a "Rez Zone" there where you can rez objects...
Our machine is running Windows Server 2008 R2 64bit.
I have tried numerous configurations, but currently I am running:
[BulletSim]
UseSeparatePhysicsThread = true
TerrainImplementation=0
AvatarToAvatarCollisionsByDefault = true
[XEngine]
;# {Enabled} {} {Enable the XEngine scripting engine?} {true false} true
;; Enable this engine in this OpenSim instance
Enabled = true
;; How many threads to keep alive even if nothing is happening
MinThreads = 2
;; How many threads to start at maximum load
MaxThreads = 250
IdleTimeout = 60
Priority = "AboveNormal"
MaxScriptEventQueue = 300
ThreadStackSize = 462144
AppDomainLoading = true
ScriptStopStrategy = co-op
DeleteScriptsOnStartup = true
DefaultCompileLanguage = "lsl"
AllowedCompilers = "lsl"
CompileWithDebugInformation = true
AllowMODFunctions = false
AllowOSFunctions = true
Allow_osTeleportAgent = true
Allow_osAvatarName2Key = true
Allow_osNpcSay = true
AllowLightShareFunctions = false
OSFunctionThreatLevel = VeryHigh
EventLimit = 1800
KillTimedOutScripts = true
ScriptDelayFactor = 1.0
ScriptDistanceLimitFactor = 1.0
NotecardLineReadCharsMax = 255
SensorMaxRange = 96.0
SensorMaxResults = 16
DisableUndergroundMovement = true
ScriptEnginesPath = "ScriptEngines"
Thank you.
_______________________________________________
Opensim-dev mailing list
[email protected]
http://opensimulator.org/cgi-bin/mailman/listinfo/opensim-dev
_______________________________________________
Opensim-dev mailing list
[email protected]
http://opensimulator.org/cgi-bin/mailman/listinfo/opensim-dev
--
Justin Clark-Casey (justincc)
OSVW Consulting
http://justincc.org
http://twitter.com/justincc
_______________________________________________
Opensim-dev mailing list
[email protected]
http://opensimulator.org/cgi-bin/mailman/listinfo/opensim-dev
