Jupyterlab application is running on kubernetes.
and when i run big model like LLM, it uses more cpu than it has granted.
for example, cpu request/limit is 10/10 but when running LLM model container is using 15core, and then CPU throttling happens. and i can’t use jupyterlab because it is unresponsive. even if i kill LLM model process it still can’t access.
is there any option to restrict cpu? or why it is unresponsive even i killed LLM process