Mysterious Intermittent no video

[Larry_Y_]

We have two identical MCU servers setup. One is a local server and the other is remote.

Chrome version: 51 64-bit

Local MCU: Open the basic example multiple times and the video always shows.

Remote MCU: Open the basic example multiple times and the video shows for the first couple of times and then either customized background only or black background only but no video, close browser and wait for a while. Repeat the above. It's inconsistent. The no video case is about 25% of the time.

Browser log for no video with customized background:

INFO: stream published: 553153661778196700 
INFO: subscribed: 57b4e9864927ebae40d29d5d
The end

Browser log for no video with black background:

INFO: stream published: 478118335362523800
INFO: subscribed: 57b4e9864927ebae40d29d5d
INFO: stream added: 478118335362523800
INFO: stream 478118335362523800 is from me; will not be subscribed.
INFO: stream 57b4e9864927ebae40d29d5d VideoLayoutChanged
The end

Server log:

2016-09-06 23:22:16.413  - INFO: VideoNode - Video engine init OK
2016-09-06 23:22:23,214  - WARN: VideoMixerWrap - Current Regions in Layout:1

2016-09-06 23:22:23,247  - INFO: woogeen.RawTransport - Increasing the buffer size: 3506
2016-09-06 23:22:34,379  - INFO: woogeen.RawTransport - Increasing the buffer size: 5511
2016-09-06 23:22:42,297  - INFO: woogeen.RawTransport - Increasing the buffer size: 7610
video-19104946-a6f9-e926-c043-2eee7329be87.log (END)

It seems that somehow the server has a glitch and does not process full-cycle requests or something happens during the network trip.

Any hint would be appreciated.

If needed, we can open up the server for you to try.

Best,

Larry

[Larry_Y_]

Sure, we will keep it open for the rest of the week.

As to side effects, we noticed two minor items: 1) It takes longer to see the mixed video. But I think it's OK and might be unavoidable; 2) When user1 joins, he sees himself. Then user 2 joins, they still just see one participant in the mixed video. This happened only once.  It might be because that the MCU did not catch up yet. We were not able to reproduce for a second time though. The two minor items might be related to each other.

Overall, the patch is a success. 

[Lei_Z_Intel1]

Hi, Larry

We have completed the problem double check and you can revoke the machine access permission. Thanks again. Please try again to see whether there is any difference between old and new versions on the mix stream show speed. We will check the code logic on the occasional publish fail issue you mentioned, if so will provide patch in 3.2.1 release.

Best wishes,

Zhai Lei 

[Larry_Y_]

Hi Lei,

The patch seems to be reliable and solid. The speed is good too. Thank you for the good work. Looking forward to the patch release.

Cheers,

L