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
It would be useful if Argo Rollouts was able to handle a group of Deployment objects, or have a separate custom resource to handle different variations of a Deployment.
Currently the Rollout resource only handles ReplicaSets that matches only one Deployment object, but different implementations could vary in nodeSelector / nodeAffinity.
Use Cases
In Google Kubernetes Engine, it's possible to save costs by using dual deployments of standard and spot virtual machines. Spot VMs are cheaper and can be used in greater numbers to increase compute capacity at low cost.
A Kubernetes Service is going to be pointing at all times to two different ReplicaSet resources (one spot, one standard).
The dual ReplicaSets resources are going to have the same images, volumes, labels, etc.
The dual ReplicaSets are going to differ in nodeAffinity / nodeSelector labels, as one is going to match specifically spot VMs while the other is going to use standard VMs.
The text was updated successfully, but these errors were encountered:
Summary
It would be useful if Argo Rollouts was able to handle a group of
Deployment
objects, or have a separate custom resource to handle different variations of aDeployment
.Currently the
Rollout
resource only handlesReplicaSet
s that matches only oneDeployment
object, but different implementations could vary in nodeSelector / nodeAffinity.Use Cases
In Google Kubernetes Engine, it's possible to save costs by using dual deployments of standard and spot virtual machines. Spot VMs are cheaper and can be used in greater numbers to increase compute capacity at low cost.
Service
is going to be pointing at all times to two differentReplicaSet
resources (one spot, one standard).ReplicaSets
resources are going to have the same images, volumes, labels, etc.ReplicaSets
are going to differ in nodeAffinity / nodeSelector labels, as one is going to match specifically spot VMs while the other is going to use standard VMs.The text was updated successfully, but these errors were encountered: