MantisBT - opensim
View Issue Details
0008506opensim[REGION] OpenSim Corepublic2019-03-23 06:192019-03-30 21:45
mewtwo0641 
 
normalmajorrandom
newopen 
 
 
Grid (Multiple Regions per Sim)
ubODE
.NET / Windows64
None
All
0008506: Some coalesced items in inventory refuse to rez
In v0.9.x and master some coalesced items in inventory completely refuse to rez in world. You'll hear the building/rezzing sound effect as you attempt to rez it but no objects appear and no scripts are loaded. This issue does not happen on v0.8.x.

The items affected appear to be random (Although the affected items remain consistent between viewer relog and restarts of OpenSim) and affect previously existing inventory items as well as sometimes coalesced items taken to inventory within the same OpenSim session.

If each piece is taken up individually it appears to rez with no problems, albeit one piece at a time. It is when an object is taken up as a group into a coalesced object that the issue can appear.

I can't figure out how to reproduce this and, in fact, attempts to reproduce it from scratch will probably not be successful unless you're just really (un)lucky.

I took up 2 groups of objects from my region today and one was able to rez and the other was not able to rez. The only thing I can think of is that the group that failed to rez had some mesh objects and the group that had successful rez only had prims/sculpts. I did test the theory and tried taking up a different group of objects that had a mixture of prims, sculpts, and meshes and see if it would fail to rez from inventory, but it rezzed out fine, so back to square one in not knowing how this is being triggered unfortunately...
No tags attached.
Issue History
2019-03-23 06:19mewtwo0641New Issue
2019-03-23 06:26mewtwo0641Description Updatedbug_revision_view_page.php?rev_id=7894#r7894
2019-03-23 08:35UbitUmarovNote Added: 0034960
2019-03-23 15:15BillBlightNote Added: 0034965
2019-03-23 15:15BillBlightSeverityblock => major
2019-03-23 15:15BillBlightNote Edited: 0034965bug_revision_view_page.php?bugnote_id=34965#r7896
2019-03-23 16:38UbitUmarovNote Added: 0034967
2019-03-23 18:50mewtwo0641Note Added: 0034968
2019-03-23 18:50mewtwo0641Note Edited: 0034968bug_revision_view_page.php?bugnote_id=34968#r7898
2019-03-23 18:51mewtwo0641Note Edited: 0034968bug_revision_view_page.php?bugnote_id=34968#r7899
2019-03-23 18:55mewtwo0641Note Edited: 0034968bug_revision_view_page.php?bugnote_id=34968#r7900
2019-03-30 12:41BillBlightNote Added: 0034997
2019-03-30 12:42BillBlightNote Edited: 0034997bug_revision_view_page.php?bugnote_id=34997#r7917
2019-03-30 19:01mewtwo0641Note Added: 0035001
2019-03-30 19:03mewtwo0641Note Edited: 0035001bug_revision_view_page.php?bugnote_id=35001#r7919
2019-03-30 19:50mewtwo0641Note Added: 0035002
2019-03-30 19:55mewtwo0641Note Edited: 0035002bug_revision_view_page.php?bugnote_id=35002#r7921
2019-03-30 19:59mewtwo0641Note Edited: 0035002bug_revision_view_page.php?bugnote_id=35002#r7922
2019-03-30 21:38mewtwo0641Note Added: 0035003
2019-03-30 21:39mewtwo0641Note Edited: 0035003bug_revision_view_page.php?bugnote_id=35003#r7924
2019-03-30 21:45BillBlightNote Added: 0035004
2019-03-30 21:48mewtwo0641Note Edited: 0035003bug_revision_view_page.php?bugnote_id=35003#r7925

Notes
(0034960)
UbitUmarov   
2019-03-23 08:35   
made a few tests with more and 256 basic prims plus meshes from mantis 8505

and all did rez fine :(
(0034965)
BillBlight   
2019-03-23 15:15   
Changed to Major from Block, a Block should indicate something that stops the simulator completely.

Cannot repo this either ...

(0034967)
UbitUmarov   
2019-03-23 16:38   
( commit 33986aea5e2ced1be33ef3b8d5c361742b17c427 incorrectly points to this mantis issue, it does refer to 8505)
(0034968)
mewtwo0641   
2019-03-23 18:50   
(edited on: 2019-03-23 18:55)
It gets even more odd... I created a new blank region to try to isolate the cause of this issue and found that on the new region I could rez the coalesced object that I took up from my other region just fine, but when I go back to the original region that I took it from and try to rez it, it won't rez.

I took the opportunity to rez it out on the new region and then take it back to inventory (with a different object as the last selected object this time) and then go back to the original region and try to rez the newly taken item but it still would not rez.

Wondering if maybe sometimes when a group of objects is taken up; not all of the references to those objects being on the region are being removed from database, perhaps due to lag hiccup while taking the objects and it misses its chance to tell the database to also remove the references. So when attempting to rez it back out the server sees that there appear to be references to those objects already on the land and it says "Wait a minute... These exact object references are already rezzed... I don't need to do anything further."? I am probably looking in the wrong direction on that but it's the only thing I can think of at the moment as an idea of where to start; it would explain why I would be able to create a new region and be able to rez it there and still not be able to rez it in the region it came from.

@Bill - Ahh, apologies. My thinking was that not being able to access inventory items was something more than major. But obviously not as major as the sim stopping completely ;)

@Ubit - I believe this is going to be one of those hard to reproduce scenarios, but when it does trigger, it is bad because it makes whatever was taken up impossible to use.

(0034997)
BillBlight   
2019-03-30 12:41   
(edited on: 2019-03-30 12:42)
@mewtwo0641 just for completeness, have you cleared the caches in the region where stuff will not rez? Just a thought ...

(I know does not really make sense, but seen stranger)

(0035001)
mewtwo0641   
2019-03-30 19:01   
(edited on: 2019-03-30 19:03)
@Bill - I just tried clearing viewer cache and region caches. Directories deleted were assetcache, maptiles, MeshCache, and ScriptEngines.

I know maptiles and ScriptEngines aren't cache but as you said, stranger things.

After doing that I was able to rez the problematic object one time; but was never able to rez it again successfully after that; even after shutting OS down, deleting the caches again, and starting OS back up... So I don't know if deleting the caches had any effect at all and that one time successful rez might have been a fluke.

Going back to my other region (Same OpenSim.exe instance as the afflicted region) I can rez the affected object(s) as many times as I like with no issue.

(0035002)
mewtwo0641   
2019-03-30 19:50   
(edited on: 2019-03-30 19:59)
I feel extremely embarrassed. I think I have figured out why it wouldn't rez. It seems OpenSim prevents rezzing an object if any part of the object would cross over the region border where there is no adjacent region, that is, just water void.

With coalesced objects I am assuming that effect is more likely to happen, especially with large builds (Large in size, not necessarily large in prims/number of objects), because there is more than one loose object which means more than one object that have their own root prim. If any of those individual objects' root would overlap the region border then the entire thing would fail to rez.

The reason I was so confused about this though is because on OS 0.8.X it would just let me rez it, region border overlap or not. Back on OS master I was further confused because on one region I could rez it and on a different region I could not. I now realize the difference between the two regions was WHERE I was attempting to rez the object... On the one region it kept failing, I was attempting to rez it more in the corner of the region because that is the general area I wanted it to go (and because there was really no other room to try to rez it due to the fact I had built out a small "city" of sorts on the remainder of the region). On the region I was able to rez it just for testing I was pretty much rezzing it dead center of the region. I tested this theory on both of the regions and sure enough, on both regions I was able to rez it out all I wanted provided I stayed clear of the region borders, but when I tried to rez near the corner of the regions it would not rez. In my previous note about it only working one time it turns out that I just happened to rez it in just the right spot near the corner where the whole thing would actually fit within the region's borders, but it was so close to where I was originally trying to rez it out and failed that I didn't make that correlation immediately.

I understand this is probably a safety feature to try to prevent stuff from getting stuck in the void, but would it be possible to put in some sort of notification of failure to rez and the reason why? Perhaps an alert to the viewer and/or a warning in the console? Just so that it's more clear to the user why their object isn't being rezzed.

I apologize for the confusion in this mantis... This one really threw me for a loop all things considered admittedly.

(0035003)
mewtwo0641   
2019-03-30 21:38   
(edited on: 2019-03-30 21:48)
Upon further testing I have found out that my previous note does not only apply to regions with water void surround; it applies even if there are regions surrounding the region to rez the the object in as well. I made a test 3 X 3 grid of regions like this:

[ ][ ][ ]
[ ][X][ ]
[ ][ ][ ]

and went to the center region where the 'X' is and attempted to rez a coalesced object near the region border and it still failed.

This has made rezzing a different larger coalesced object of mine pretty much impossible on OS master on a normal sized region because it spans greater than 256 meters... I had to create a large var region to accommodate rezzing it out and then take it back into smaller chunks so I can rez it out in multiple pieces on a normal sized region.

Should rezzing fail in the instance given that there is actually land that can support holding the objects and the user rezzing the object has build rights?

(0035004)
BillBlight   
2019-03-30 21:45   
coalesced objects can have multiple roots, only link sets have a single root, so it is quite possible that coalesced objects are too much outside the region.