It has been a pain for us to manually update the resourceQuota in vc table when there is quota shifting from a vc to another. This PR
exposes a rest api /ResourceQuota for admin to update resourceQuota in vc. Resources for GPU sku will be set proportionally to requested GPU quota, while resources for CPU sku wikk be set proportionally to requested CPU quota. The restful API server will invalidate the local cache afterwards, saving us the extra step to restart it. @Gerhut need your help to add a dashboard page for cluster admin to manage VC quota.
add a command in ctl.py to query/update resource quota for vc:
It has been a pain for us to manually update the
resourceQuota
invc
table when there is quota shifting from a vc to another. This PR/ResourceQuota
for admin to updateresourceQuota
invc
. Resources for GPU sku will be set proportionally to requested GPU quota, while resources for CPU sku wikk be set proportionally to requested CPU quota. The restful API server will invalidate the local cache afterwards, saving us the extra step to restart it. @Gerhut need your help to add a dashboard page for cluster admin to manage VC quota.ctl.py
to query/update resource quota for vc:$ ./ctl.py -x quota show # detail view (GPU, GPU memory, CPU, memory) gpu:
vc Standard_ND24rs
platform 12 vc1 0
gpu_memory (Gi):
vc Standard_ND24rs
platform 288 vc1 0
cpu:
vc Standard_ND24rs
platform 72 vc1 0
memory (Gi):
vc Standard_ND24rs
platform 1344 vc1 0
$ ./ctl.py quota update --vc_name platform --sku Standard_ND24s --quota 13