Skip to content

chore: Advance libsveltos pin and drift-detection-manager image tag - #1916

Merged
gianlucam76 merged 1 commit into
projectsveltos:mainfrom
gianlucam76:dd
Aug 14, 2026
Merged

chore: Advance libsveltos pin and drift-detection-manager image tag#1916
gianlucam76 merged 1 commit into
projectsveltos:mainfrom
gianlucam76:dd

Conversation

@gianlucam76

Copy link
Copy Markdown
Member

This will pick:

  1. An optimization on the CustomResourceDefinition watcher: replaces the SharedIndexInformer used to watch CRDs with a lean client-go Reflector that forwards each Add/Update/Delete straight to the handler and discards the object, instead of retaining every CRD in an indexed cache for the process lifetime. Reduces memory consumption on every Sveltos component, including addon-controller, starting that watcher.
  2. The same optimization applied to drift-detection-manager's per-GVK resource watchers: replaces the DynamicSharedInformerFactory backed informer used to watch tracked resources with the same reflector/discarding-store approach. Nothing there ever read the cached store back (drift checks always re fetch the resource live to evaluate if a configuration drift happened), so it was pure overhead, retained per watched GVK for as long as the watcher stayed active. Reduces drift-detection-manager memory on clusters tracking many resources across many GVKs.

This will pick:
1. An optimization on the CustomResourceDefinition watcher:
   replaces the SharedIndexInformer used to watch CRDs with a lean
   client-go Reflector that forwards each Add/Update/Delete straight to
   the handler and discards the object, instead of retaining every CRD
   in an indexed cache for the process lifetime. Reduces memory consumption
   on every Sveltos component, including addon-controller, starting that
   watcher.
2. The same optimization applied to drift-detection-manager's per-GVK
   resource watchers: replaces the DynamicSharedInformerFactory backed
   informer used to watch tracked resources with the same
   reflector/discarding-store approach. Nothing there ever read the
   cached store back (drift checks always re fetch the resource live to
   evaluate if a configuration drift happened), so it was pure overhead,
   retained per watched GVK for as long as the watcher stayed active.
   Reduces drift-detection-manager memory on clusters tracking many
   resources across many GVKs.
@gianlucam76
gianlucam76 merged commit 3495f86 into projectsveltos:main Aug 14, 2026
19 of 20 checks passed
@gianlucam76
gianlucam76 deleted the dd branch August 14, 2026 17:25
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant