[Opensim-users] Scripts sometimes do not load on start up?

[Chris]

Alright, I posted a mantis along with the results of 'show threads' here

http://opensimulator.org/mantis/view.php?id=6651

Thanks Justin!

On 5/22/2013 3:26 PM, Justin Clark-Casey wrote:
> I think this would be worth opening a Mantis for.  In particular, I'd 
> be interested to see the results of "show threads" when this happens.  
> A VM thread dump would be ideal but that's a trickier proposition on 
> Windows.
>
> On 15/05/13 00:32, Chris wrote:
>> Whoops... Hit the send button a bit too early :)
>>
>> In addition to the testing done in the previous post, I found that it 
>> seems like the only option that seems to more
>> consistently let all regions load their scripts is to set 
>> DeleteScriptsOnStartup back to true; in this case both my
>> systems load all scripts on their regions on both databases... But it 
>> significantly slows down startup due to having to
>> compile all the scripts which is the reason I usually set it to false.
>>
>> On 5/14/2013 6:13 PM, Chris wrote:
>>> Hi Justin,
>>>
>>> I'm running 11 regions in Standalone (No HG) configuration. 5 of 
>>> those regions do not have any scripts to be loaded on
>>> them and very few prims to load.
>>>
>>> I disabled 5 regions, leaving 6 regions to load at start up, 
>>> restarted OpenSim a few times (I always shut OpenSim down
>>> completely and then restart it from the exe file rather than use the 
>>> restart command), but found that it doesn't seem
>>> to make a difference in script start up.
>>>
>>> Set async_call_method = UnsafeQueueUserWorkItem in OpenSim.ini and 
>>> restarted OpenSim a few times. From what I can see,
>>> setting this seems to make the issue appear more frequently, and on 
>>> more regions than with using default setting;
>>> though this could just be coincidence as I set it back to 
>>> SmartThreadPool and tried again with similar results.
>>>
>>> Set TrustBinaries = false with no noticable improvement.
>>>
>>> Set ScriptDelayFactor = 1.0 with no noticable improvement.
>>>
>>> I also tried one extra thing, setting AppDomainLoading = false, but 
>>> this didn't seem to have noticeable improvement.
>>>
>>> The issue seems to show up randomly and will pick random region(s) 
>>> to not load scripts for on start up. It's a
>>> slightly older commit but I don't see the issue showing up in 
>>> 3611d33 (r/22387) with all my regions enabled and the
>>> script settings reported earlier set.
>>>
>>> Ran these tests on two different machines with two different 
>>> databases with different content to load.
>>>
>>> On 5/14/2013 4:25 PM, Justin Clark-Casey wrote:
>>>> I've not heard any other reports of this yet though this doesn't 
>>>> necessarily mean anything.
>>>>
>>>> How many regions are we talking about here?  What happens if you 
>>>> run fewer regions on one machine?  Does specifying
>>>>
>>>> async_call_method = UnsafeQueueUserWorkItem
>>>>
>>>> in [Startup] on OpenSim.ini make any difference? (this uses the 
>>>> system pool for threads rather than a separate
>>>> SmartThreadPool instance)?
>>>>
>>>> Also, does setting TrustBinaries = false make any difference? This 
>>>> is the one relatively unusual setting you have
>>>> below.  You could also try specifying a non zero ScriptDelayFactor 
>>>> in case some script is managing to hog CPU.
>>>>
>>>> On 11/05/13 04:15, Chris wrote:
>>>>> I was noticing while testing master that sometimes when starting 
>>>>> OpenSim up that it is refusing to load scripts for
>>>>> some
>>>>> or all regions at start up. Once every thing is loaded up and the 
>>>>> console is ready, running the 'login status' command
>>>>> it shows that some regions are enabled login, and some aren't. In 
>>>>> order to login the 'login enable' command must be ran
>>>>> first. Once logged in it seems that no script will run on the 
>>>>> regions that have not had their scripts loaded at start
>>>>> up, including attachments, or newly rezzed objects. Had a closer 
>>>>> look at the console out put and was noticing that
>>>>> there
>>>>> are a lot of warnings dealing with objects I have rezzed in world 
>>>>> that are mesh similar to below:
>>>>>
>>>>> WARN  - OpenSim.Region.Physics.Meshing.Meshmerizer [MESH]: No 
>>>>> recognized physics mesh found in mesh asset for
>>>>> chair/c861affd-17b4-4b3c-9535-f1986ab394a9
>>>>>
>>>>> I found that if I tracked these objects down and removed them from 
>>>>> the region and then restarted OpenSim, there's a bit
>>>>> better chance for scripts to be loaded properly. If I put the 
>>>>> objects back out again and restart OpenSim then it starts
>>>>> to randomly not want to load scripts in random regions at start 
>>>>> up. Setting DeleteScriptsOnStartup = true in
>>>>> OpenSim.ini
>>>>> or deleteing the ScriptEngines folder prior to start up also seems 
>>>>> to give better success at having scripts load
>>>>> properly. It might be just my imagination and coincidence that it 
>>>>> is working in this way but it is my observation.
>>>>>
>>>>> Noticed that the console seems to hang for a bit (a couple 
>>>>> seconds) before outputting warnings on these objects with
>>>>> high CPU use. I'm just speculating on this since I'm not entirely 
>>>>> knowledgeable on how things work but... could there
>>>>> possibly be some kind of race condition somewhere preventing 
>>>>> scripts from loading properly in this case?
>>>>>
>>>>> A couple of my regions have 0 scripts on them; not sure if it 
>>>>> matters or not.
>>>>>
>>>>> I normally run OpenSim with these script related settings:
>>>>> AppDomainLoading = false
>>>>> DeleteScriptsOnStartup = false
>>>>> EventLimit = 500
>>>>> ScriptDelayFactor = 0.0
>>>>> AllowScriptCrossing = true
>>>>> TrustBinaries = true
>>>>>
>>>>> Is anyone else seeing this?
>>>>>
>>>>> Thanks!
>>>>>
>>>>
>>>>
>>>
>>>
>>
>>
>
>


-- 
OpenSim: 10 Region Standalone on 0.7.6 Dev
Physics: Open Dynamics Engine
OS: Windows 7 (x64)
CPU: AMD Phenom II X4 840 3.2 GHz
Memory: 11 GB DDR3
Database: MySQL 5.1.63 (x64)