Blocked GPU resources

[Ifigeneia Apostolopoulou]

+1 I have also observed that (and had to individually send email to users
to kill useless processes that were devouring computing).

On Thu, Dec 16, 2021 at 1:16 PM Benedikt Boecking <boecking at andrew.cmu.edu>
wrote:

> Hello everyone,
>
> Many lab members currently have idle processes on our GPU servers that are
> hindering usage of the gpus by others. This can happen for many reasons,
> including forgetting about jupyter notebooks or errors that stall a script
> without all threads being closed. While the processes are idle, they can
> still hog GPU and server memory.
>
> I would like to ask you to please log on to the GPU servers you have used
> to check if this is the case for any of your processes. You can check this
> on a server by using nvidia-smi to see which processes are running on which
> GPU, and you can use htop and filter for your username or process ID to see
> if the process IDs are yours in case the process name shown on nvidia-smi
> doesn’t give it away immediately.
>
> Thanks in advance for your collaboration!
>
> Best,
> Ben
>
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/autonlab-users/attachments/20211216/9d32a158/attachment.html>