recvmsg encountered uncommon error: Message too long

General
1

image
image1920×675 107 KB

When I use the Flower framework and the AlexNet model for federated learning, I encounter the error message “recvmsg encountered uncommon error: Message too long”. However, the AlexNet model has only 136.78MB of parameters when trained on the CIFAR-10 dataset, which is well below the 512MB limit. Why is this happening?

2

Hi @liujiaxi,

Thanks for posting your question here. If you try to run your code using a more lightweight model, does the problem still persist?

3

I ran the code using the LeNet model, and there is no such problem.