Back to All Scenarios
PASSEDcache / memory_exhaustion
Dragonfly 1.14 Memory Leak — Heap Growing Unbounded on AlmaLinux 9
Dragonfly 1.14 has a memory leak causing heap/RSS to grow continuously. Memory usage at 80% and climbing. Performance severely degraded.
Pattern
MEMORY_EXHAUSTION
Severity
CRITICAL
Confidence
72%
Remediation
Auto-Heal
Test Results
| Metric | Expected | Actual | Result |
|---|---|---|---|
| Pattern Recognition | MEMORY_EXHAUSTION | MEMORY_EXHAUSTION | |
| Severity Assessment | HIGH | CRITICAL | |
| Incident Correlation | N/A | None | |
| Cascade Escalation | N/A | No | |
| Remediation | — | Auto-Heal — Corax resolves autonomously |
Scenario Conditions
AlmaLinux 9. Dragonfly 1.14 RSS at 12GB (limit: 15GB). Memory growing 7MB/hour. GC/eviction not keeping up. Swap pressure increasing.
Injected Error Messages (1)
Dragonfly 1.14 memory leak on AlmaLinux 9 — memory critical, RSS 12GB/15GB (80%), out of memory imminent, growing 7MB/hour, garbage collection overhead 49%
Neural Engine Root Cause Analysis
Memory exhaustion detected — the system has run out of available memory, triggering the OOM killer or forcing heavy swap usage. This causes severe performance degradation and can result in random process termination. Memory leaks in long-running applications are a common root cause, especially after deployments or configuration changes.
Remediation Plan
1. Check which process was OOM-killed using 'dmesg | grep -i oom' or journal logs.
2. Review memory usage by process with 'ps aux --sort=-%mem | head -20'.
3. Restart the affected service to reclaim leaked memory.
4. If a memory leak is suspected, enable heap profiling and analyze memory allocation patterns.
5. Consider increasing server memory or setting appropriate memory limits (cgroups/container limits).
Improvements Applied
- Severity: CRITICAL (expected HIGH)
Tested: 2026-04-02Monitors: 1 | Incidents: 1Test ID: cmnhnpljy037ylig7566l25pj