You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi, I am still using Kubernetes version 1.11 in the production environment.
After research, I found that the HAMi project's GPU scheduling and compute/memory isolation modules meet our needs. I would like to inquire if it is possible to tailor the HAMi project to exclude the hami-scheduler for global unified scheduling, and instead only handle the GPU scheduling part( combined the vgpu-scheduler-extender with our kube-scheduler 1.11). If feasible, I hope to get your response and some suggestions for furthe custom development.
Waiting for your reply.
The text was updated successfully, but these errors were encountered:
Currently, HAMi extends the second kube-scheduler through the scheduler-extender method. If you do not need an additional scheduler, you can also add a sidecar extension to the existing kube-scheduler.
Currently, HAMi extends the second kube-scheduler through the scheduler-extender method. If you do not need an additional scheduler, you can also add a sidecar extension to the existing kube-scheduler.
1. Issue or feature description
Hi, I am still using Kubernetes version 1.11 in the production environment.
After research, I found that the HAMi project's GPU scheduling and compute/memory isolation modules meet our needs. I would like to inquire if it is possible to tailor the HAMi project to exclude the hami-scheduler for global unified scheduling, and instead only handle the GPU scheduling part( combined the vgpu-scheduler-extender with our kube-scheduler 1.11). If feasible, I hope to get your response and some suggestions for furthe custom development.
Waiting for your reply.
The text was updated successfully, but these errors were encountered: