| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
| |
|
|
|
|
|
|
|
|
|
| |
* This will show the packets waiting in each queue for each client logged into a region server
* These are displayed using 'show stats' on the region command line
* This is in pursuit of a memory leak.
* This will require a prebuild
|
| |
|
|
|
|
|
|
| |
Updated OpenSim.ini.example
|
|
|
|
| |
instances to prevent event race conditions
|
| |
|
|
|
|
| |
clients), problem is that we don't seem to be persisting PrimitiveBaseShape.State in the datastore, so after restart of server, they all turn back into pine trees.
|
| |
|
| |
|
|
|
|
|
|
|
| |
regardless of number of regions is now 3. This will save you around 33 threads for a normal 3x3 region server.
But, this is totally completely untested. So it probably won't work for another patch or five.
|
|
|
|
|
|
|
|
| |
per region.
Total so far: 2 threads less per region
Note: Currently causes delay in load/unload of scripts
|
| |
|
|
|
|
|
|
|
|
| |
* Speeding up ScriptEngine shutdown
* Sharing threads so that minimum total thread count for any amount of regions will be 2. (1 maintenance, 1 script execution)
You can choose more script exec threads if you want of course.
In this commit: Sharing maintenance thread between all regions.
|
|
|
|
| |
recommended format.
|
| |
|
|
|
|
|
|
|
| |
sends a message back to ClientView to close it a second time. (yes we call close twice because there are times when the client closes with a logout and disconnects immediately which causes the UDP server to send a close request to the client thread)
* This update might make things better, it might expose another issue.
|
|
|
|
| |
don't try to enumerate dead null ODECharacter objects when things get *really* slow.
|
| |
|
|
|
|
| |
compiled in DEBUG mode)... Its ugly and even requires a separate thread to track the treads, but it will be very valuable in debugging.
|
| |
|
|
|
|
| |
an asset back from the cache
|
|
|
|
|
|
|
| |
* Resolve a bad logic bug in AssetCache.GetAsset()
* This may make some asset related things work better (possibly getting main map images will now be improved).
|
| |
|
|
|
|
|
|
|
|
|
| |
data from the asset server
* This should stop the constant increase in the download requests statistics
* If you see stat numbers for download requests which are far from what you'd expect, please report
|
|
|
|
|
| |
threads console command will list all threads. This + yesterdays naming threads patch will give a good overview of what threads we have running.
|
| |
|
| |
|
|
|
|
| |
the source of the null on multicast_void: error?
|
| |
|
|
|
|
|
|
|
| |
* This fixes some of the 'runaway downloads' problem but not all of it
* Also fix up logging messages so texture requests are reported as such rather than as assets
|
|
|
|
| |
in timed scripts...
|
|
|
|
|
|
| |
* turned script asset fetching asynchronous
|
|
|
|
| |
AbortXfer packets. And the "saving data" message in the client on logout seems to have gone. (So that message was all my fault.)
|
|
|
|
| |
they help with the xfer/grey avatar problems.
|
| |
|
|
|
|
| |
Groups Module, made it shared to save on threads.
|
|
|
|
|
|
|
|
| |
scene)
* Fixed ClickAction situation on the same note (properties shouldn't cause big changes)
* Added some more debug output to AssetCache
|
|
|
|
|
|
| |
* Untangled a tangly shutdown loop for the ScenePresence.
* Suggested to the Garbage Collector that this may be a good time to >.>, <.< *gasp* collect the memory.
|
|
|
|
|
|
|
|
|
| |
themselves for updates looking for changes. This runs 10 times a second.
* Set the massively slow UpdateEntities method to run every 2 seconds instead of 10 times a second. This method runs through *all* of the entities can calls the virtual update().
* Documented some of the code in the scene.Update method.
|
| |
|
|
|
|
|
|
|
|
| |
* The reason why pending downloads tick ever upwards is because missing assets are never signalled to the TextureSender
* Rectifying this is not straightfoward, but this will constitute the next patch.
* This does not explain the memory leak.
|
|
|
|
| |
makes it easier to pick out a restart of OpenSim in an appended log file
|
|
|
|
|
|
|
|
| |
This patch implements the llSendRemoteData command and fixes mantis 552,
and possibly 586.
|
|
|
|
|
|
|
| |
with a triangle hollow, Cube with a cylinder hollow, cylinder with a triangle hollow, etc.
* More work needs to be done here as several of the profileshape/hollow combinations are rotationally offset from where they render in the client.
|
|
|
|
| |
threads so it will be easier to debug.
|
| |
|
|
|
|
| |
object types in the drop down on the object tab. Positive tapers are slightly incorrect(prim sinks into ground a tiny bit). Everything else that's supported works as expected. Hollow, cut, negative tapers, top shear.
|
|
|
|
| |
ASCII!
|
|
|
|
|
|
|
|
| |
Patch provided by Openlifegrid.com
Adds locks around clientCircuits and clientCircuits_reverse Adds existance check on adding to clientCircuits for clients resending UseCircuit packet.
Adds locks around Clientmanager.m_clients for add/remove/tryGet calls
|