Description
Please investigate the issue with the recording BOT user using the key: 'recording_bot202505261133289386d6'. This user is unable to render another participant’s camera feed on the canvas — the video appears completely black. However, other participants’ camera feeds still render correctly on the video player.
Additionally, there seems to be no audio output, so the user may not hear any other speakers.
Could you check what’s causing this? Is it possibly related to the auto-play-audio-failed error as mentioned in this Zoom Dev Forum issue, or could there be another underlying cause?
This is the session ID: Gj5dxAvDQNi3ANn7dqMJ3w==
Thank you in advance!
Which Web Video SDK version?
2.1.0
Video SDK Code Snippets
# Other users start camera using this func
let videoOptions = {
cameraId: cameraId,
originalRatio: window.isMobile
};
if (stream.isSupportVirtualBackground() && virtualBackground) {
videoOptions.virtualBackground = { imageUrl: virtualBackground };
}
await stream.startVideo(videoOptions);
# Bot user render other user camera on canvas
const canvasElement = document.getElementById(ViewIdentifiers.PIN_CANVAS_VIEW);
canvasElement.setAttribute('user-id', userId);
await stream.renderVideo(canvasElement, userId, canvasElement.width, canvasElement.height, 0, 0, quality);
# Bot user starts audio with speaker `true` to hear other speakers
await stream.startAudio({ speakerOnly: true });
Device (please complete the following information):
Since we still don’t have access to the detailed logs, based on our current analysis, the no-audio issue is still likely due to the browser’s auto-play policy.
As for the video rendering issue, the root cause was mentioned in another thread—it’s due to the virtual machine failing to decode video frames. We expect this to be fixed in the next version.
Do you know when the next version is scheduled to be released?
Specifying video_webrtc_mode: 1 in the JWT explicitly enables WebRTC video, which can indeed help with video rendering on virtual machines that lack physical GPUs
Are there any other ways—such as upgrading or modifying the EC2 instance configuration (used by the BOT user)—to reduce the frequency of this bug on the BOT client side?
Do you know when the next version is scheduled to be released?
The issue occurs on a virtual machine using a headless browser, which is an uncommon setup.
We need more time to investigate and find a solution. It’s expected to be addressed in the next release (end of June), but we cannot guarantee it at this point.
Are there any other ways—such as upgrading or modifying the EC2 instance configuration (used by the BOT user)—to reduce the frequency of this bug on the BOT client side?
If we don’t enable WebRTC and use the video-player instead of rendering via canvas, and use attachVideo() instead of renderVideo() to render the video — could this help resolve or at least reduce the frequency of the bug?
The issue where the canvas stays completely black and appears frozen when rendering video using canvas only started happening after we migrated to Video SDK version 2.1.0. It did not occur when we were using version 1.11.10.
Would downgrading back to version 1.11.10 resolve the issue? Is this a known regression in 2.1.0?