You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Ok I know why if I have more invoked lambda even if I removed the the status progression.
When I chunked my video into N, Each chunk will invoked 2 lambda functions (render and main), the render will stream to the main lambda the data and then the main is creating the output.
One issue with this, the main started at the same time as the chunk function, so the billed duration is * 2 for N chunks.
Main:
2024-10-06T16:31:27.012Z - START RequestId: xx
2024-10-06T16:32:06.806Z - INFO Received video chunk for chunk 0
[39 seconds without doing anything] = lost $ and you multiply this * chunks_count
It's because Remotion is designed to have only one Main function that received streams from all the others rendered lambda.
But as I mentioned, preres profile concat is a very long process.
So my idea it's to have the possibility when calling renderMediaOnLambda to have a noStream option, so we don't invoked the main function that combined the stream. We just stored the output into s3 output with the render function.
What do you think <@701030996430159872>
The text was updated successfully, but these errors were encountered:
From message posted on Discord by 0x_master
Ok I know why if I have more invoked lambda even if I removed the the status progression.
When I chunked my video into N, Each chunk will invoked 2 lambda functions (render and main), the render will stream to the main lambda the data and then the main is creating the output.
One issue with this, the main started at the same time as the chunk function, so the billed duration is * 2 for N chunks.
Main:
2024-10-06T16:31:27.012Z - START RequestId: xx
2024-10-06T16:32:06.806Z - INFO Received video chunk for chunk 0
[39 seconds without doing anything] = lost $ and you multiply this * chunks_count
It's because Remotion is designed to have only one Main function that received streams from all the others rendered lambda.
But as I mentioned, preres profile concat is a very long process.
So my idea it's to have the possibility when calling renderMediaOnLambda to have a noStream option, so we don't invoked the main function that combined the stream. We just stored the output into s3 output with the render function.
What do you think <@701030996430159872>
The text was updated successfully, but these errors were encountered: