Closed felipejfc closed 7 years ago
The informer has a resync period set to 60 seconds, which should trigger the OnUpdate handler even without any actual changes. This should catch any missed events on nodes. https://github.com/ottoyiu/k8s-ec2-srcdst/blob/master/pkg/controller/srcdst_controller.go#L46
as documented here: https://github.com/kubernetes/client-go/blob/450baa5d60f8d6a251c7682cb6f86e939b750b2d/tools/cache/controller.go#L270
Have you run into any scenarios where the controller/cache informer missed an update?
No, I haven't... Just to be sure as I'm running a very important production cluster with it now
is it possible that the cache informer loses the update on some node somehow?
if so, I think that we should also have a periodic check on all nodes for self healing them...
what do you think @ottoyiu ?