Jicofo fails to recognize Jibri's successful launch the first time

I have a custom AMI with Jibri set up. Whenever I launch an instance (AWS EC2 - c5a.xlarge) from the AMI for the first time, Jicofo fails to recognize Jibri’s successful launch and says:

Unexpected response to start request: null

On the frontend, I get “All recorders are busy”. However, Jibri does work and record the meeting. In addition, everything is fine from the next time.

I have read other threads regarding the issue but none of them helped. I am planning to run a bunch of EC2s using AWS Auto Scaling Groups, thus it is critical to get them working from the first time.

Thanks so much.

3 Likes

I also faced this issue recently, and still searching threads. Unable to find any answers. It would be great if anyone could help me as well.

You may try to decrease the jibri resolution to 720p

Thanks for the reply, @emrah. It is already in 720p. Though, I would be interested in understanding how does that affect the set up? :slight_smile:

Facing the same issue in AWS ec2.

This may be related with the initial run of Chrome in jibri instance

Hmm. I did try a suggestion from another thread to start a chrome with icewm on startup and close it post 30 seconds. But that didn’t work out either.

Do you have any customization to configure Jibri? For example a script which run before jibri and configures it

Nothing destructive. I have a docker swarm setup with an overlay network. I just have edited the .yml files to use that overlay network.

I didn’t try the dockerized setup, so no idea for this environment

Ah, I see. Folks with non-Docker setup have also reported the issue in other threads.

Same issue here!

Earlier, I had the Jibri image built already in the AMI. I am not sure why, but it is now working when I build and run the image on boot.