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
What would you like to be added:
Currently rescheduleCriticalSeconds in Adaptive scheduling workloadspread has a max limit of 295 seconds with 2 subsets.
Additional subsets will bring down this allowed rescheduleCriticalSeconds .
Feature/RFE is to not put a hard limit and let user choose what they want.
Or if that is not architecturally possible, allow a limit which is more than 295. Something atleast 15 minutes
Invalid value: 180: rescheduleCriticalSeconds < 0 or rescheduleCriticalSeconds > 98 is not permitted
I have 4 subsets .
Adding each subset lower the allowed value of reschedulecriticalseconds.
Can we allow user to specify a reasonable time such as 5 minutes.? Cloud Kubernetes deployments take atleast 2 minutes to spin up a node in response to an autoscaler event
Why is this needed:
Scenario:
I have multiple node pools in Azure AKS. It takes 3 minutes for a node to be spun up and get added to node pool inresponse to an autoscaler event.
If I create a workload spread with 4 subsets, the maxallowed is 98 seconds which is not enough to bring a new kubernets node into the cluster.
admission webhook "vworkloadspread.kb.io" denied the request: spec.scheduleStrategy.adaptive.rescheduleCriticalSeconds: Invalid value: 295: rescheduleCriticalSeconds < 0 or rescheduleCriticalSeconds > 98 is not permitted
apiVersion: apps.kruise.io/v1alpha1
kind: WorkloadSpread
metadata:
name: frontend-workloadspread
spec:
targetRef:
apiVersion: apps/v1
kind: Deployment
name: frontend
subsets:
- name: spotvmmodel1
requiredNodeSelectorTerm:
matchExpressions:
- key: type
operator: In
values:
- spotvmmodel1
maxReplicas:
- name: spotvmmodel2
requiredNodeSelectorTerm:
matchExpressions:
- key: type
operator: In
values:
- spotvmmodel2
maxReplicas:
- name: spotvmmodel3
requiredNodeSelectorTerm:
matchExpressions:
- key: type
operator: In
values:
- spotvmmodel3
maxReplicas:
- name: spotvmmodel4
requiredNodeSelectorTerm:
matchExpressions:
- key: type
operator: In
values:
- spotvmmodel4
scheduleStrategy:
type: Adaptive
adaptive:
rescheduleCriticalSeconds: 295
The text was updated successfully, but these errors were encountered:
What would you like to be added:
Currently rescheduleCriticalSeconds in Adaptive scheduling workloadspread has a max limit of 295 seconds with 2 subsets.
Additional subsets will bring down this allowed rescheduleCriticalSeconds .
Feature/RFE is to not put a hard limit and let user choose what they want.
Or if that is not architecturally possible, allow a limit which is more than 295. Something atleast 15 minutes
scheduleStrategy:
type: Adaptive
adaptive:
rescheduleCriticalSeconds: 180
Invalid value: 180: rescheduleCriticalSeconds < 0 or rescheduleCriticalSeconds > 98 is not permitted
I have 4 subsets .
Adding each subset lower the allowed value of reschedulecriticalseconds.
Can we allow user to specify a reasonable time such as 5 minutes.? Cloud Kubernetes deployments take atleast 2 minutes to spin up a node in response to an autoscaler event
Why is this needed:
Scenario:
I have multiple node pools in Azure AKS. It takes 3 minutes for a node to be spun up and get added to node pool inresponse to an autoscaler event.
If I create a workload spread with 4 subsets, the maxallowed is 98 seconds which is not enough to bring a new kubernets node into the cluster.
admission webhook "vworkloadspread.kb.io" denied the request: spec.scheduleStrategy.adaptive.rescheduleCriticalSeconds: Invalid value: 295: rescheduleCriticalSeconds < 0 or rescheduleCriticalSeconds > 98 is not permitted
apiVersion: apps.kruise.io/v1alpha1
kind: WorkloadSpread
metadata:
name: frontend-workloadspread
spec:
targetRef:
apiVersion: apps/v1
kind: Deployment
name: frontend
subsets:
- name: spotvmmodel1
requiredNodeSelectorTerm:
matchExpressions:
- key: type
operator: In
values:
- spotvmmodel1
maxReplicas:
- name: spotvmmodel2
requiredNodeSelectorTerm:
matchExpressions:
- key: type
operator: In
values:
- spotvmmodel2
maxReplicas:
- name: spotvmmodel3
requiredNodeSelectorTerm:
matchExpressions:
- key: type
operator: In
values:
- spotvmmodel3
maxReplicas:
- name: spotvmmodel4
requiredNodeSelectorTerm:
matchExpressions:
- key: type
operator: In
values:
- spotvmmodel4
scheduleStrategy:
type: Adaptive
adaptive:
rescheduleCriticalSeconds: 295
The text was updated successfully, but these errors were encountered: