Back to All Scenarios
PASSEDcloud / azure_vm
Azure: Azure VM Disk Swap — Temp Disk Data Lost
An Azure VM was migrated during Azure maintenance. The temporary disk (D:) was reassigned and all data on it was lost, including TempDB and page files.
Pattern
DATABASE_EVENT
Expected: AZURE_VM_FAILURE
Severity
HIGH
Confidence
68%
Remediation
Auto-Heal
Test Results
| Metric | Expected | Actual | Result |
|---|---|---|---|
| Pattern Recognition | AZURE_VM_FAILURE | DATABASE_EVENT | |
| Severity Assessment | CRITICAL | HIGH | |
| Incident Correlation | N/A | None | |
| Cascade Escalation | N/A | No | |
| Remediation | — | Auto-Heal — Corax resolves autonomously |
Scenario Conditions
Azure VM Standard_D4s_v3. Temporary disk D: used for SQL TempDB. Azure host maintenance triggered live migration. Temp disk wiped. SQL Server TempDB missing.
Injected Error Messages (1)
Azure VM temp disk wiped — VM 'sql-prod-01' migrated during Azure maintenance, temporary disk D: data lost, SQL Server TempDB was on D: and is now missing, SQL Server in recovery mode
Neural Engine Root Cause Analysis
Database infrastructure event detected — the connection pool may be exhausted preventing new connections, replication lag is growing between primary and replica, deadlocks are occurring between competing transactions, or slow queries are degrading overall database performance. Database issues cascade to affect all applications and services that depend on the database.
Remediation Plan
1. For connection pool exhaustion, check current connections with 'SHOW PROCESSLIST' or 'pg_stat_activity' and identify idle/stuck connections.
2. For replication lag, check replica I/O and SQL thread status and identify long-running transactions on the primary.
3. For deadlocks, review the deadlock graph (InnoDB: 'SHOW ENGINE INNODB STATUS', Postgres: check pg_locks) and optimize transaction ordering.
4. For slow queries, enable and review the slow query log, add missing indexes, and optimize query plans with EXPLAIN.
5. Consider scaling read replicas or implementing connection pooling (PgBouncer/ProxySQL) if connection limits are consistently hit.
Improvements Applied
- Pattern classified as DATABASE_EVENT (expected AZURE_VM_FAILURE)
- Severity: HIGH (expected CRITICAL)
Tested: 2026-04-02Monitors: 1 | Incidents: 1Test ID: cmnhnr8vc09bmlig72m7hx9wy