clientcmd.BuildConfigFromFlags and kubernetes.NewForConfig were called on every tick of the 1-minute loop, incurring unnecessary file I/O and allocations each cycle.

Changes

• Clientset lifecycle: A buildClientset closure builds the REST config and clientset once before the loop. Inside the loop it is only invoked when clientset == nil.
• Failure-triggered recreation: After a failed Nodes().Get() call, clientset is set to nil so the next tick transparently rebuilds it — covering kubeconfig rotation and transient API server unavailability.
• Goroutine resilience: Previous return statements on per-tick errors replaced with continue; transient failures no longer kill the goroutine.
• Initial failure visibility: If the pre-loop build fails, an info log is emitted to make the first-tick retry observable.

// Built once; reused every tick
buildClientset()

for {
    select {
    case <-ticker.C:
        if clientset == nil && !buildClientset() {
            continue
        }
        node, err := clientset.CoreV1().Nodes().Get(...)
        if err != nil {
            clientset = nil  // force rebuild next tick
            continue
        }
        ...
    }
}

📍 Connect Copilot coding agent with Jira, Azure Boards or Linear to delegate work to Copilot in one click without leaving your project management tool.