Applier Kubernetes Events

Kubernetes Events emitted by the StormForge Applier: what they mean, when they fire, and existing tooling support

The StormForge Applier emits standard Kubernetes Events. If your team uses kubectl, Lens, k9s, or a cluster event exporter to track cluster activity, you can track Applier activity without any additional configuration.

For push-based notifications when the Applier applies a recommendation, see Configure applier alerts.

View events

Use kubectl to query events on any Applier-managed workload:

# Events on a specific workload
kubectl get events -n <namespace> --field-selector involvedObject.name=<workload-name>

# Filter by event reason (substitute any reason from the reference below)
kubectl get events -n <namespace> --field-selector reason=RecommendationApplied

# Watch for a specific reason in real time
kubectl get events -n <namespace> -w --field-selector reason=RecommendationApplied

Events also appear in kubectl describe deployment <name> (or statefulset, daemonset, etc.) under the Events: section.

Event output

When auto-apply applies a recommendation to a Deployment:

$ kubectl get events -n hipster \
    --field-selector involvedObject.name=cartservice,reason=RecommendationApplied

LAST SEEN   TYPE     REASON                  OBJECT                   MESSAGE
2m          Normal   RecommendationApplied   Deployment/cartservice   StormForge successfully applied recommendation patches

Each event includes:

Field	Description
`type`	`Normal` or `Warning`
`reason`	A stable string you can key on for dashboards and alerts (see reference below)
`message`	A human-readable description
`involvedObject`	The workload (Deployment, StatefulSet, etc.) or HPA the event describes

Event reference

Recommendation lifecycle

Events are attached to the affected workload.

Reason	Type	Message	When it fires
`RecommendationApplied`	Normal	StormForge successfully applied recommendation patches	Patches applied to the workload
`RecommendationApplyFailed`	Warning	StormForge failed to apply recommendation patches	Patch application encountered an error
`RecommendationExpired`	Warning	StormForge timed out waiting to process this recommendation	Processing timeout exceeded before patches could be applied

Resource changes

Events are attached to the patched object, for example, the updated Deployment or StatefulSet.

Reason	Type	Message	When it fires
`WorkloadResourcesChanged`	Normal	StormForge applied resource changes (: summary if available)	Resource changes recorded by the Applier; the optional summary lists changed fields, for example: `server cpu-requests 100m→50m, server memory-requests 256Mi→128Mi`

WorkloadResourcesChanged fires once per patched object. RecommendationApplied fires once per recommendation after all objects are patched. Both fire for the same apply operation.

Rollout validation

After applying patches, the Applier validates that the workload becomes healthy. These events fire on the affected workload when validation does not succeed.

Reason	Type	Message	When it fires
`RolloutValidationFailed`	Warning	StormForge rollout validation failed: details	Workload unhealthy after patches applied; rollback follows unless disabled
`RolloutValidationTimedOut`	Warning	StormForge timed out waiting for the rollout to become healthy	Validation window exceeded; patches are left in place
`RolloutValidationAbandoned`	Warning	StormForge rollout validation failed with an unrecoverable error	Unrecoverable error during validation

Rollback

Events are attached to the affected workload.

Reason	Type	Message	When it fires
`RecommendationRolledBack`	Warning	StormForge rolled back recommendation patch: details	Patches reverted after failed rollout validation
`RecommendationRollbackFailed`	Warning	StormForge failed to roll back patches after validation failure	Rollback attempt also failed

In-place pod resize

Events are attached to the affected workload. These fire only when the Applier attempts to resize pods without triggering a rollout.

Reason	Type	Message	When it fires
`InPlaceResizeSucceeded`	Normal	StormForge successfully resized all pods in place	All pods in the workload resized without a rollout
`InPlaceResizeFailed`	Warning	StormForge in-place pod resize failed, falling back to regular patches	In-place resize failed; Applier retries with a standard patch

Drift reconciliation

The Applier watches workloads and HPAs it manages. If an external controller modifies fields the Applier owns, the Applier detects the change and re-applies the last recommendation. Workload and HPA targets share the same reason strings but produce distinct messages.

Workload

Reason	Type	Message	When it fires
`RecommendationDriftDetected`	Warning	StormForge detected drift: fields modified by manager	An external manager modified workload fields owned by the Applier
`RecommendationDriftReconciled`	Normal	StormForge reconciled drift to match the applied recommendation	Applied recommendation re-applied to the workload after drift

HPA

Reason	Type	Message	When it fires
`RecommendationDriftDetected`	Warning	StormForge detected HPA drift: fields modified by manager	An external manager modified HPA fields owned by the Applier
`RecommendationDriftReconciled`	Normal	StormForge reconciled HPA drift to match the applied recommendation	Applied recommendation re-applied to the HPA after drift

Operational

Reason	Type	Message	Target	When it fires
`SuspendState`	Warning	applier started with suspend=value	Applier Deployment	Applier started with a non-default suspend configuration

Use events with your existing tooling

These events work with any standard Kubernetes tooling: dashboards such as Lens and k9s, cluster event exporters (Datadog, New Relic, Dynatrace), Prometheus via an event exporter, or custom automation keyed on the stable reason field.

Last modified June 9, 2026