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
At what point do we start to hit Azure throttling?
At what point do we start to have issues with the Kubernetes control plane (# of CRs)
Ideally we could define some upper bounds. Things to consider here are:
Spike in load on pod restart as it re-hydrates lists and sends requests to Azure.
Resource shape/size. K8s controlplane will have much less issue with tons of ResourceGroups (small) than tons of ManagedClusters (large).
Secrets integration: Some resources watch secrets, which means the operator is subscribed to secrets events, so possibly a ton of (unrelated) secrets activity could cause issues. Doubly so if the secrets activity is actually on secrets we're paying attention to.
The text was updated successfully, but these errors were encountered:
We're still interested in doing this, although our current experience suggests that we run into Azure scalability limits (request throttling) before we run into Kubernetes scaling limits on most things.
The obvious exception being CRD size driving APIServer memory usage
We should strive to answer two questions:
Ideally we could define some upper bounds. Things to consider here are:
The text was updated successfully, but these errors were encountered: