fix hami bug when vgpu-number no set #3867
Conversation
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: The full list of commands accepted by this bot can be found here.
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
volcano-sh-bot
added
the
size/XS
Signed-off-by: weapons97 <[email protected]> Signed-off-by: weipeng <[email protected]>
|
/assign @k82cn |
|
Yes, i think sooner or later, we need to add a mutatingwebhookConfiguration for volcano.sh/vgpu-memory, manually add a volcano.sh/vgpu-number: 1 to these tasks. Since it doesn't conflict with this fix, so i'll vote approve |
|
/lgtm |
| return true | ||
| } | ||
| _, ok = container.Resources.Limits[VolcanoVGPUNumber] | ||
| _, ok := container.Resources.Limits[VolcanoVGPUNumber] |
There was a problem hiding this comment.
What if user just set VolcanoVGPUNumber but not set VolcanoVGPUMemory？
There was a problem hiding this comment.
it will try to find an unused GPU, and use that exclusively, if such can't be find, then it will be stuck in pending state.
|
Please add ut to cover all possible cases. |
In the Hami project volcano-vgpu-device-plugin , if the user does not set volcano.sh/vgpu-number, it will cause a bug where the plugin does not respond to container resource mounts. Therefore, the check in checkVGPUResourcesInPod needs to be modified.
related issue:
Project-HAMi/volcano-vgpu-device-plugin#40