Fix label generation for node discovery #1116
Merged
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
A set of labels are used to find all the nodes of a cluster of ingress controllers. The controller instance running the code takes its own set of labels and use it to find the others via apiserver. The set of labels has, however, some labels that uniquely identify a pod, like pod-template-hash created by the ReplicaSet, making the node discovery to discover only one node, the one being running. On shutdowns, this makes the controller to incorrectly think it is running alone, leading to remove the status of all the ingress resources. Removing all the well-known label keys that uniquely identifies a pod solves this issue.