🔒 Paid
Primary Database Failover
The primary database's health is degrading and will become fully unavailable if nothing changes. Application services read and write through a DB Proxy sitting in front of a Primary and a Replica. Order and account data must remain correct above all else: a write that is silently lost, or a write accepted by a node that should no longer be authoritative, is a critical data-integrity incident, not a performance blip. Restore write availability to a durable, sustainable state.
- Difficulty
- Staff
- Expected Time
- ~15 min
- Category
- Databases & Storage
Skills
- Reasoning about availability vs. correctness tradeoffs during failover
- Evaluating replication lag before promoting a replica
- Recognizing split-primary/concurrent-writer risk
- Deciding when fencing is required to make a cutover safe
System Architecture
- Application Services
- DB Proxy
- Primary
- Replica
This is a Pro scenario. Upgrade to Pro to unlock it.
Upgrade to ProBack to Incidents