DaemonSet availability alerts
A DaemonSet runs a Pod on each matching node. When a Pod is unavailable, a node, scheduling, image, or application problem may be preventing the DaemonSet from doing its job. kwatch alerts after the condition has lasted long enough to avoid brief rollout noise.
What kwatch detects
| Setting | Default | Purpose |
|---|---|---|
daemonSetMonitor.enabled | true | Watch unavailable DaemonSet Pods. |
daemonSetMonitor.sustainedMinutes | 5 | Wait before alerting on an unavailable Pod. |
The sustained window helps avoid notifications for short rolling updates or temporary node changes. Increasing it delays real alerts too.
Investigate the missing Pod
kubectl describe daemonset <name> -n <namespace>
kubectl get pods -n <namespace> -o wide
kubectl get nodes
Compare desired and available Pods, then inspect the affected node and Pod Events. If a node is NotReady, investigate the node first. If only one DaemonSet Pod fails, check its image, scheduling constraints, and logs.
Configure the monitor
Use the interactive manager to change settings. The equivalent configuration fragment is:
daemonSetMonitor:
enabled: true
sustainedMinutes: 5
Run kwatch lint after a change. See node and workload coverage
for related signals.