Reconciliation of Network for a hibernated Shoot takes more than 30s #191
Labels
area/networking
Networking related
kind/bug
Bug
lifecycle/rotten
Nobody worked on this for 12 months (final aging stage)
How to categorize this issue?
/area networking
/kind bug
What happened:
Recently we had the following regression in gardener/gardener - ref gardener/gardener#8005. In short, several steps had too low timeouts. For example the wait step for the Network to be reconciled had timeout of 30s.
With this issue we noticed that for hibernated cilium Shoots, the Network cannot be reconciled for 30s.
Explanation:
Reconcile
func, networking-cilium callsgetIPAMMode
(refgardener-extension-networking-cilium/pkg/controller/actuator_reconcile.go
Lines 149 to 152 in d4869c4
getCiliumConfigMap
func is being called (refgardener-extension-networking-cilium/pkg/controller/actuator_reconcile.go
Lines 191 to 194 in d4869c4
getCiliumConfigMap
tries to build a Shoot client and get a ConfigMap from the Shoot (gardener-extension-networking-cilium/pkg/controller/actuator_reconcile.go
Lines 180 to 188 in d4869c4
I locally measured the execution time of the
getCiliumConfigMap
func with this diff:Logs for non-hibernated Shoot:
Logs for hibernated Shoot:
You can see that the
getCiliumConfigMap
func itself takes more than 30s when the Shoot is hibernated (no kube-apiserver Pod running).What you expected to happen:
The Network reconciliation to be fast when the Shoot is hibernated.
How to reproduce it (as minimally and precisely as possible):
See above.
Anything else we need to know?:
Environment:
kubectl version
): Shoot K8s version is v1.25.9The text was updated successfully, but these errors were encountered: