Repository navigation
Handlers that enqueue all ClusterRoles / all LogicalClusters on the shard #4396
Description
Every LogicalCluster add lists and enqueues all ClusterRoles on the shard (core/replicateclusterrole → EnqueueClusterRoles), and every APIExport or WorkspaceType event with resources/default bindings enqueues all LogicalClusters on the shard, bound or not (defaultapibindinglifecycle, as its comment says).
With 1e5 workspaces and a few ClusterRoles each, shard start and workspace creation become quadratic, and a single platform APIExport update requeues the whole shard.
Suggestion: enqueue only the new cluster's ClusterRoles; index LogicalClusters by the types/exports they actually use.
Activity
PragalvaXFREZ commentedon Oct 13, 2026
Hi, can I work on the ClusterRole part of #4396? I think we can scope the requeue to the affected logical cluster and add tests for it first. We could handle the APIExport/WorkspaceType changes in a follow-up.