Skip to content

Kubernetes watch can create duplicate informers after watch resource changes #2255

Description

@adityathebe

config-db currently caches Kubernetes shared informers by scraper ID, full watch resource list, and auth fingerprint. In a long-running process, if a scraper's watch resources change, the cache key changes and a new dynamic informer factory is created. Any resource that exists in both the old and new watch lists is not stopped by the existing cleanup logic, so the same scraper/cluster can end up with multiple running informers for the same GroupVersionKind. This can also happen when the same watch resources are supplied in a different order, because the watch-list string is order-sensitive.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions