[gianlucam76]

When deploying an EventTrigger to a cluster failed, or when the ClusterProfile instances could not be generated from an EventReport, the failure was only recorded in the status (status.clusterInfo[].failureMessage of the EventTrigger and status.failureMessage of the EventReport) and in the logs. Nothing showed in kubectl describe or in the events of the namespace.

event-manager now records a Warning event for both. It uses the EventTriggerFailed reason defined in libsveltos lib/eventreasons.

When	Regarding	Action	Message
Deploying an EventTrigger to a cluster fails	the EventTrigger, with the cluster as the related object	Deploying	Deploying to cluster Sveltos mgmt/prod-1 failed: <error>

The EventTrigger is cluster scoped, so its events are in the default namespace. The cluster is the related object because the events recorder merges events with the same reason, action and regarding object and keeps only the first note: without it, the failures of one EventTrigger in two clusters would be a single event.

An event is recorded when the failure message is new or different from the one saved before, not on every retry:

• Per cluster, the previous failure message of the cluster is kept in status.clusterInfo through the Provisioning step of a retry. A retry which fails the same way records nothing, a different error does.
• For an EventReport, the event is recorded in setEventReportFailureMessage, which already skips the write when the message is unchanged. Nothing is recorded when the failure is cleared, or when only AgentFailureMessage changed.

No event is recorded for expected states: a paused cluster, a cluster which is not ready or down, a cluster whose agent is not healthy (sveltoscluster-manager records that one), and a deployment which is still in progress. Those return before a deployment result exists.