Back to All Scenarios
PASSEDcloud / gcp_gke
GCP: GKE Node Auto-Repair Triggered — Node Unresponsive
GKE detected an unhealthy node and triggered auto-repair, which involves draining and recreating the node. Disruption to pods during repair.
Pattern
UNKNOWN
Expected: GCP_GKE_REPAIR
Severity
MEDIUM
Confidence
68%
Remediation
Auto-Heal
Test Results
| Metric | Expected | Actual | Result |
|---|---|---|---|
| Pattern Recognition | GCP_GKE_REPAIR | UNKNOWN | |
| Severity Assessment | HIGH | MEDIUM | |
| Incident Correlation | N/A | None | |
| Cascade Escalation | N/A | No | |
| Remediation | — | Auto-Heal — Corax resolves autonomously |
Scenario Conditions
GCP GKE cluster 'prod'. Node 'gke-prod-pool-abc123' unresponsive for 10 minutes. Auto-repair triggered. 20 pods being evicted. PodDisruptionBudgets may be violated.
Injected Error Messages (1)
GKE auto-repair — node 'gke-prod-pool-abc123' marked unhealthy, auto-repair draining node, 20 pods being evicted, PDB violations possible, temporary capacity reduction
Neural Engine Root Cause Analysis
Unrecognized error pattern — this may be a new type of failure not yet cataloged.
Remediation Plan
Trigger autonomous healing to diagnose and fix. Manual investigation may be needed if healing cannot resolve.
Improvements Applied
- Pattern unrecognized — consider adding keywords for GCP_GKE_REPAIR
- Severity: MEDIUM (expected HIGH)
Tested: 2026-04-02Monitors: 1 | Incidents: 1Test ID: cmnhnr8vc09brlig7wka26zce