New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
torch causes Fatal Python error: Floating point exception #1198
Comments
This seems to be related to PyTorch. Hard crashes of PyTorch usually involve a bug in some instruction set of the CPU. Can you give me more information what kind of CPU you use and if there is maybe any virtualization involved? |
It is an older computer but should have sufficient memory and storage. I'm running Fedora CoreOS on bare metal so no virtualization. I didn't see a particular CPU requirement in the pyTorch documentation. Any idea what it needs? How do I bypass or disable pyTorch? uname -a cat /proc/cpuinfo
|
We upgraded to PyTorch 2.3, maybe this got fixed in that release :) |
🐛 Bug Report
📝 Description of issue:
The log is filled with python exception traces like the below. I'm scanning in tens of thousands of photos on a fresh Docker install.
🔁 How can we reproduce it:
Unsure. This happened on a fresh install. I reproduced it by deleting all the librephotos and database folders and running again. I'm running on podman instead of docker but the web interface is working well and I can see that it has found my photos. I don't think the torch library should cause the librephotos job to crash like this. Does it need some exception handling to fail more gracefully?
It's certainly possible this is an artifact of using podman. Here is the podman kube file I'm using with podman play kube (note that in podman Pods, all containers share an IP address and localhost):
Please provide additional information:
The text was updated successfully, but these errors were encountered: