webrtc node crash during test

[qwu16]

Following steps below:

1. All OWT modules deployed in device 1 and configure webrtc agent region to BJ
2. User B publish stream and start analytics with algorithm a1
3. User A join with region SH and subscribe B's analyzed stream
4. Launched webrtc agent with region SH and analytics agent with algorithm a2 on device 2
5. Call restful API to analyze B's stream with algorithm a2
6. User A unsub B's analyzed stream a1 and subscribe B's analyzed stream a2
7. stop analytics agent on device 2
8. User A unsub B's analyzed stream a2 and subscribe B's analyzed stream a1

repeat step 4-8 several times and webrtc node crashed once with following crash log:

node: ../deps/uv/src/unix/core.c:280: uv__finish_close: Assertion `0' failed.

[starwarfan]

Can you provide core dump?

[zhouhuirun2015]

I faced the same problem.
Steps:
mobile(android/ios) browser refreshed frequency,the server cpu will up to max limitation,about 85% and NEVER calm down.
Logs:

VCMFrameEncoder said no valid buffer, Ok, change the buffer max num from 3 to 100.Try steps again.

Logs(Time 21:00):
webrtc-agent

webrtc-node

video-node

[zhouhuirun2015]

server:4.3, 16 cores 32GB, all in one cluster
client:4.3
mobile browser client use the auto mix stream like example.
send camera res: auto
receive stream res:1080P

[cisco66666]

@zhouhuirun2015 I use 4.3 MCU, a user logs in through chrome, pub a stream, sub a forward stream and a mix stream, repeatedly refresh the page many times, the cpu has been in a stable state, did not reproduce your situation

[zhouhuirun2015]

@cisco66666 poor server hardware

[starwarfan]

We're investigating the CPU issue recently. Which process consumes most CPU? Could you provide detail CPU load in your scenario using command like "top -c"?