aboutsummaryrefslogtreecommitdiffstatshomepage
path: root/OpenSim/Region/ClientStack/LindenUDP/LLUDPServer.cs (unfollow)
Commit message (Collapse)AuthorFilesLines
2010-11-08Formatting cleanup.Jeff Ames1-2/+2
2010-10-30Set async_packet_handling = true by defaultJustin Clark-Casey (justincc)1-1/+1
Setting this to true avoids a 500ms or so client freeze when the LLUDP server thread is taken up with processing a UseCircuitCode packet synchronously. Extensive testing on Wright Plaza appeared to show no bad effects and this seems to reduce login lag considerably. Of course, a lot of login lag is still coming from other sources.
2010-10-29Add number of ms it takes to complete UseCircuitCode packet handling to log ↵Justin Clark-Casey (justincc)1-2/+8
for diagnostics
2010-10-22Revert "Merge remote branch 'otakup0pe/mantis5110'"Justin Clark-Casey (justincc)1-1/+1
This reverts commit 21187f459ea2ae590dda4249fa15ebf116d04fe0, reversing changes made to 8f34e46d7449be1c29419a232a8f7f1e5918f03c.
2010-10-21* change the data exchanged within hypergrid transactionsJonathan Freedman1-1/+1
2010-09-17Send KillPackets on the Task queue rather than the State queueJustin Clark-Casey (justincc)1-0/+4
Object updates are sent on the task queue. It's possible for an object update to be placed on the client queue before a kill packet comes along. The kill packet would then be placed on the state queue and possibly get sent before the update If the update gets sent afterwards then client get undeletable no owner objects until relog Placing the kills in the task queue should mean that they are received after updates. The kill record prevents subsequent updates getting on the queue Comments state that updates are sent via the state queue but this isn't true. If this was the case this problem might not exist.
2010-09-13Add client name to packet resend log messages to make them a bit more ↵Justin Clark-Casey (justincc)1-3/+7
informative
2010-08-08In my crusade against facelights, I am striking the killing blow. Add aMelanie1-0/+5
DisableFacelights option to OpenSim.ini to finally kill those immersion- breaking, silly vanity lights that destroy nighttime RP. Girls, you look just fine without them. Guys, you too. Thank you. Melanie has left the building.
2010-05-12* Initial commit of the slimupdates2 rewrite. This pass maintains the ↵unknown1-14/+4
original behavior of avatar update sending and has a simplified set of IClientAPI methods for sending avatar/prim updates
2010-04-02Don't append ACKs to zerocoded packets. Although this should fine in theory, ↵John Hurliman1-15/+20
I'm seeing the viewer ignore or fail to parse ACKs appended to our zerocoded packets. This should cut down on viewer->sim resend traffic
2010-03-16Removed the unused use_async_when_possible config variableJohn Hurliman1-1/+1
2010-02-23Streamlined error logging for malformed packets and fixed a bug when ↵John Hurliman1-4/+3
printing the hex dump
2010-01-29* HGGridConnector is no longer necessary.Diva Canto1-0/+3
* Handle logout properly. This needed an addition to IClientAPI, because of how the logout packet is currently being handled -- the agent is being removed from the scene before the different event handlers are executed, which is broken.
2010-01-28HG 1.5 is in place. Tested in standalone only.Diva Canto1-20/+0
2010-01-18Go Home works. With security!!Diva Canto1-0/+20
2009-12-13* Add some glue that allows LocalPacketHandlers to decide if they want the ↵Teravus Ovares (Dan Olivares)1-1/+1
packet to be processed asynchronously or not. * Make several packets not asynchronous (such as AgentUpdate). In theory, all fast returning packet handling methods should not be asynchronous. Ones that wait on an external resource or a long held lock, should be asynchronous.
2009-11-27* Re-enable lightweight packet tracking stats on a 3000 ms interval.Teravus Ovares (Dan Olivares)1-0/+9
2009-11-05Applying #4332, optional packet statistics loggingJohn Hurliman1-0/+111
2009-11-04Refine oversized packet handling as per jhurliman. Reallocate the buffer toMelanie1-4/+7
actual packet size only for oversized packets.
2009-11-04Add some length to the backet buffer for packet sending so oversizeMelanie1-1/+1
inventory packets don't make us barf
2009-10-30Make the default and max RTO configurable int he linden client stackMelanie1-1/+7
2009-10-30* Disables automatic packet splitting on AvatarGroupsReply packets. This ↵John Hurliman1-2/+2
packet is a mess and shouldn't be used at all (in favor of the event queue message) * Clean up the way we send AvatarGroupsReply packets, including clamping the group name and group title
2009-10-30* Moving parcel media and avatar update packets from the unthrottled ↵John Hurliman1-4/+8
category to task * Fixing a bug where the max burst rate for the state category was being set as unlimited, causing connections to child agents to saturate bandwidth * Upped the example default drip rates to 1000 bytes/sec, the minimum granularity for the token buckets
2009-10-30* Fixes issue #4329 "llDialog fails silently" by updating OpenMetaverse.dllJohn Hurliman1-4/+4
* Prints a warning for any future packet splitting failures
2009-10-29* Commented out two noisy debug lines in the LLUDP serverJohn Hurliman1-2/+2
* Misc. cleanup in ScenePresence.HandleAgentUpdate()
2009-10-28Modified visibilities of properties and methods. This is so that SL client ↵Mikko Pallari1-5/+5
based clients can use UDP server that is inherited from LLUDPServer.
2009-10-27Experimental test to rate limit the incoming packet handler and try to ↵John Hurliman1-0/+8
always leave a worker thread available for other tasks
2009-10-26Removing the ClientManager reference from IScene and hiding it entirely ↵John Hurliman1-7/+7
inside Scene as an implementation detail. This will reduce programming error and make it easier to refactor the avatar vs client vs presence mess later on
2009-10-26* Switched all operations on the list of clients that could be either sync ↵John Hurliman1-2/+2
or async to use Scene.ForEachClient() instead of referencing ClientManager directly * Added a new [Startup] config option called use_async_when_possible to signal how to run operations that could be either sync or async * Changed Scene.ForEachClient to respect use_async_when_possible * Fixing a potential deadlock in Parallel.ForEach by locking on a temporary object instead of the enumerator (which may be shared across multiple invocations on ForEach). Thank you diva
2009-10-26Changed UseCircuitCode handling to be synchronous or asynchronous depending ↵John Hurliman1-10/+21
on the async_packet_handling config option, and added a debug log message when a UseCircuitCode packet is handled
2009-10-25Changing Scene.ForEachClient to use the synchronous for loop instead of ↵John Hurliman1-2/+2
Parallel. This is quite possibly the source of some deadlocking, and at the very least the synchronous version gives better stack traces * Lock the LLUDPClient RTO math * Add a helper function for backing off the RTO, and follow the optional advice in RFC 2988 to clear existing SRTT and RTTVAR values during a backoff * Removing the unused PrimitiveBaseShape.SculptImage parameter * Improved performance of SceneObjectPart instantiation * ZeroMesher now drops SculptData bytes like Meshmerizer, to allow the texture data to be GCed * Improved typecasting speed in MySQLLegacyRegionData.BuildShape() * Improved the instantiation of PrimitiveBaseShape
2009-10-23* Unregister Mono.Addins event handlers in PluginLoader.Dispose() and always ↵John Hurliman1-1/+5
handle PluginLoader with the using pattern. This freed up 121,634,796 bytes on my system * Avoid allocating an Action<IClientAPI> object every round of the OutgoingPacketHandler * Removed unnecessary semi-colon endings from OpenSim.ini.example [InterestManagement] section
2009-10-23Cleaning up OpenSim.ini.example for LLUDP. The [LLClient] section has been ↵John Hurliman1-0/+24
removed and several new parameters have been added to [ClientStack.LindenUDP]
2009-10-23Uncommented the resend log line so the previous commit can be seen in actionJohn Hurliman1-1/+1
2009-10-23* Changed the max RTO value to 60 seconds to comply with RFC 2988John Hurliman1-0/+3
* Implemented section 5.5, exponential backoff of the RTO after a resend
2009-10-23Commented out noisy debugging about resent packets (normal) and agents ↵John Hurliman1-1/+1
setting throttles (normal)
2009-10-22Implemented a Watchdog class. Do not manually create Thread objects anymore, ↵John Hurliman1-8/+11
use Watchdog.StartThread(). While your thread is running call Watchdog.UpdateThread(). When it is shutting down call Watchdog.RemoveThread(). Most of the threads in OpenSim have been updated
2009-10-21* Change the OnQueueEmpty signature to send the flags of the queues that are ↵John Hurliman1-4/+0
empty instead of firing once per empty queue * Change the OnQueueEmpty firing to use a minimum time until next fire instead of a sleep * Set OutgoingPacket.TickCount = 0 earlier to avoid extra resends when things are running slowly (inside a profiler, for example)
2009-10-21* Clarified what FireQueueEmpty is doing with a MIN_CALLBACK_MS constant and ↵John Hurliman1-4/+0
upped it to 30ms * Removed the unused PacketSent() function * Switched UnackedPacketCollection from a SortedDictionary to a Dictionary now that the sorting is no longer needed. Big performance improvement for ResendUnacked()
2009-10-21* Changed the timing calculations for sending resends/acks/pings from ↵John Hurliman1-26/+59
per-client back to per-scene * Testing a fix from Jim to make the cpu usage fix cleaner
2009-10-21Removing the Sleep(0) call from the OutgoingPacketHandlerJohn Hurliman1-6/+3
2009-10-21* Handle UseCircuitCode packets asynchronously. Adding an agent to a scene ↵John Hurliman1-8/+39
can take several seconds, and was blocking up packet handling in the meantime * Clamp retransmission timeout values between three and 10 seconds * Log outgoing time for a packet right after it is sent instead of well before * Loop through the entire UnackedPacketCollection when looking for expired packets
2009-10-21* Replaced the UnackedPacketCollection with a lockless implementation. The ↵John Hurliman1-98/+45
tiny amount of time spent in the locks turned into a lot of time when the rest of the LLUDP implementation went lockless * Changed the timer tracking numbers for each client to not have "memory". It will no longer queue up calls to functions like ResendUnacked * Reverted Jim's WaitHandle code. Although it was technically more correct, it exhibited the exact same behavior as the old code but spent more cycles. The 20ms has been replaced with the minimum amount of time before a token bucket could receive a drip, and an else { sleep(0); } was added to make sure the outgoing packet handler always yields at least a minimum amount
2009-10-21Committing Jim's optimization to replace the 20ms sleep in outgoing packet ↵John Hurliman1-28/+70
handling with an interruptible wait handle
2009-10-20* Added a sanity check to GetScriptAssemblies() and GetScriptStates() for ↵John Hurliman1-0/+2
the case where no scripting engine is enabled * Added TokenBucket.cs to OpenSim, with some fixes for setting a more accurate MaxBurst value and getting a more accurate Content value (by Drip()ing each get)
2009-10-20* Removed the throttle speed optimizations to see if it brings stability backJohn Hurliman1-63/+53
* Changed the outgoing packet handler to use a real function instead of a closure and to track time on a per-client basis instead of a global basis
2009-10-20Added try/catches in the outgoing packet handler to match the one in the ↵John Hurliman1-45/+59
incoming packet handler
2009-10-19* Optimized sending of terrain dataJohn Hurliman1-9/+14
* Send terrain data in a spiral pattern instead of a typewriter pattern (placeholder until terrain data becomes part of the interest list management) * Added a debug line when resent packets are being sent
2009-10-19* Change Util.FireAndForget to use ThreadPool.UnsafeQueueUserWorkItem(). ↵John Hurliman1-7/+12
This avoids .NET remoting and a managed->unmanaged->managed jump. Overall, a night and day performance difference * Initialize the LLClientView prim full update queue to the number of prims in the scene for a big performance boost * Reordered some comparisons on hot code paths for a minor speed boost * Removed an unnecessary call to the expensive DateTime.Now function (if you *have* to get the current time as opposed to Environment.TickCount, always use DateTime.UtcNow) * Don't fire the queue empty callback for the Resend category * Run the outgoing packet handler thread loop for each client synchronously. It seems like more time was being spent doing the execution asynchronously, and it made deadlocks very difficult to track down * Rewrote some expensive math in LandObject.cs * Optimized EntityManager to only lock on operations that need locking, and use TryGetValue() where possible * Only update the attachment database when an object is attached or detached * Other small misc. performance improvements
2009-10-18* Rewrote the methods that build ObjectUpdate and ImprovedTerseObjectUpdate ↵John Hurliman1-2/+2
packets to fill in the data more accurately and avoid allocating memory that is immediately thrown away * Changed the Send*Data structs in IClientAPI to use public readonly members instead of private members and getters * Made Parallel.ProcessorCount public * Started switching over packet building methods in LLClientView to use Util.StringToBytes[256/1024]() instead of Utils.StringToBytes() * More cleanup of the ScenePresences vs. ClientManager nightmare * ScenePresence.HandleAgentUpdate() will now time out and drop incoming AgentUpdate packets after three seconds. This fixes a deadlock on m_AgentUpdates that was blocking up the LLUDP server
generated by cgit v1.1 at 2024-12-16 13:36:40 +0000