Incidents Not Appearing

Cluster is not connected

Incidents are only created for connected clusters. Go to Clusters and verify the cluster shows Connected.

If the cluster is disconnected, see Agent not connecting.

Analysis hasn’t run yet

K8sGPT analysis runs every 30 seconds, starting shortly after the agent connects. Warning events and log errors are reported as they happen.

If you just connected the cluster, the first results usually arrive within a minute.

Condition is within the suppression window

Transient conditions are suppressed to reduce noise. If the issue resolved itself quickly (within the suppression window), no incident is created.

If the condition is ongoing and still not appearing after 10 minutes, contact support.

Condition is not covered by the agent’s detectors

DataAgent detects conditions through K8sGPT analysis, Kubernetes Warning events, pod logs, and Prometheus alerts. Application errors written to pod logs are caught by the log watcher when they match one of its known error patterns — you can add your own with logWatcher.config.customPatterns (see the Helm chart reference). Some conditions are not detectable:

• Conditions inside custom resources without a K8sGPT analyzer
• Network-level issues not visible to the Kubernetes API

The incident was auto-resolved

If the condition was previously detected but fixed, the incident may have been auto-resolved. Check the Resolved filter in the Incidents view for recent resolutions.

You are filtering it out

Check the active filters in the Incidents view. The severity filter, cluster filter, or status filter may be hiding the incident. Click Clear filters to reset.