Request has timed out in Request: method:onIceCandidate

Hi,

I have set up openVidu/Docker on a dedicated Ubuntu 18.04 instance and it works very well.
In order to run load tests, I have created a test web page through which multiple parallel connections/subscriptions to a running stream can being established.
I did run these tests on different clients in different locations in parallel. On some clients I can create up to 30 connections before the client’s CPU is maxed out (bandwidth does not seem to pose a problem though), which is fine.
But at some point (sometimes 10, sometimes 50 parallel global connections from different locations) - inevitably - following below errors appear (Chrome Dev Console), when trying to create additional connections.
I have run the tests many times, granting the dedicated Ubuntu-Server 1, 4 and 8 vCPUs. The number of allocated CPUs does not seem to make any difference.

Do you have a suggestion what might cause this issue and how it could be handled?

Many Thanks in Advance
Marcel

ERROR:Request has timed out in Request: method:onIceCandidate params:{“endpointName”:“con_NREq4SAd44”,“candidate”:“candidate:2602892196 1 udp 2113937151 85b83208-bd6e-4338-8e0e-6360c3d39e02.local 50426 typ host generation 0 ufrag 8zvZ network-cost 999”,“sdpMid”:“0”,“sdpMLineIndex”:0} request:{“jsonrpc”:“2.0”,“method”:“onIceCandidate”,“params”:{“endpointName”:“con_NREq4SAd44”,“candidate”:“candidate:2602892196 1 udp 2113937151 85b83208-bd6e-4338-8e0e-6360c3d39e02.local 50426 typ host generation 0 ufrag 8zvZ network-cost 999”,“sdpMid”:“0”,“sdpMLineIndex”:0},“id”:3}
(anonymous) @ jsonrpcclient.js:102
dispatchCallback @ index.js:343
timeout @ index.js:377
invokeTask @ zone-evergreen.js:400
runTask @ zone-evergreen.js:168
invokeTask @ zone-evergreen.js:481
ZoneTask.invoke @ zone-evergreen.js:470
timer @ zone-evergreen.js:2535
setTimeout (async)
scheduleTask @ zone-evergreen.js:2556
scheduleTask @ zone-evergreen.js:386
scheduleTask @ zone-evergreen.js:211
scheduleMacroTask @ zone-evergreen.js:234
scheduleMacroTaskWithCurrentZone @ zone-evergreen.js:1118
(anonymous) @ zone-evergreen.js:2569
proto. @ zone-evergreen.js:1433
sendRequest @ index.js:354
RpcBuilder.encode @ index.js:380
JsonRpcClient.send @ jsonrpcclient.js:99
push…/node_modules/openvidu-browser/lib/OpenVidu/OpenVidu.js.OpenVidu.sendRequest @ OpenVidu.js:803
push…/node_modules/openvidu-browser/lib/OpenVidu/Connection.js.Connection.sendIceCandidate @ Connection.js:66
WebRtcPeer.pc.onicecandidate @ WebRtcPeer.js:57

OpenViduLogger.js:56 Error sending ICE candidate: {“request”:"{“jsonrpc”:“2.0”,“method”:“onIceCandidate”,“params”:{“endpointName”:“con_NREq4SAd44”,“candidate”:“candidate:2602892196 1 udp 2113937151 85b83208-bd6e-4338-8e0e-6360c3d39e02.local 50426 typ host generation 0 ufrag 8zvZ network-cost 999”,“sdpMid”:“0”,“sdpMLineIndex”:0},“id”:3}",“requestTime”:1591771843444}

Timeouts are usually caused by a stressed CPU, as the server doesn’t have enough CPU cycles to answer petitions. If you are stating that CPU spikes occur in a non-consistent way, that could reflect some kind of issue or bug in OpenVidu or Kurento. If you have any detailed information on when this timeouts take place, please post it here so we can do further research.