I would like the ability to understand how much GPU memory an algorithm is using as it is running (per running algorithm). This ability will be used for debugging purposes.
However, under kubernetes this can only be achieved if the pod is running with a hostPID option.
Please add the ability to run with hostPID. At least for debug purposes.
Other alternatives might be hkube itself logging gpu usage per algorithm, however I would still prefer my solution as I can report other statistics along with GPU usage to monitoring systems
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.
Hi,
I would like the ability to understand how much GPU memory an algorithm is using as it is running (per running algorithm). This ability will be used for debugging purposes. However, under kubernetes this can only be achieved if the pod is running with a hostPID option. Please add the ability to run with hostPID. At least for debug purposes.
Other alternatives might be hkube itself logging gpu usage per algorithm, however I would still prefer my solution as I can report other statistics along with GPU usage to monitoring systems