On Join, no streams

[Goran_J_]

Hi there,

in the basic example, we experience very often, that if couple of users have joined in the conference (non mixed mode) and everybody see's each other, then a new user joins, in the JOIN callback, it returns 1 user (self) and 0 streams, so it's not able to subscribe to any of the existing user's streams.

Usually we need 1 or 2 page refreshes and then it works.

Do u know why and what might be causing this issue?

[Lei_Z_Intel1]

Hi, Goran

We haven't encountered this problem before. Can you please share your logs when this problem occurred?

Best wises,

Zhai Lei

[Goran_J_]

Hi Lei,

I've attached the logs. The only thing that we've noticed is that it constantly get disconnected from and reconnected to the cluster. Even though we are using single MCU instance.

Let me know if you need more information. Also if needed, we can schedule a screen sharing session.

Thanks.

/GJ

[Lei_Z_Intel1]

Which MCU version are you using?

[Goran_J_]

Latest. 3.2. UPDATE 1

[Goran_J_]

Hi Lei,

any update regarding this issue? Did you find anything that suggest why this might be happening?

Tnx.

/GJ

[Lei_Z_Intel1]

Hi, Goran

Still working on that. One question is did you see any error log in browser when the response JSON callback is not correct?

Best wishes,

Zhai Lei

[Goran_J_]

Hi Lei,

unfortunately we don't see any client side nor server side errors.

However, we have implemented a "RETRY" workaround on the client side with JS promises, that retries up to 10 times with certain delay the following three steps:

1. createToken

2. join

3. leave

if the join callback returns only 1 user and 1 stream. In other words for the first member to a conference always goes via this 10 trials and for every other new participant to the conference, usually 2-3 retrials are enough to successfully join.

Currently this works well enough for us. But if you can identify why this is happening so that we avoid using this workaround, it would be really helpful.

Thanks.

/GJ

[Lei_Z_Intel1]

Hi, Goran

Thanks for your info. We have found the clue on the issue of cluster components keeping disconnect and connect from cluster manager under some special environments and fixed internally. It will be included in next major version around early Jan. Feel free to let us know if you need a patch to check whether that helps your case.

Best wishes,

Zhai Lei

[Goran_J_]

Hi Lei,

after R3.3. this really seems to fix our problem. Great work and thanks again!

/GJ

[Lei_Z_Intel1]

Nice to hear that. Enjoy!