Objective.
Kill the edge proxy that carries customer traffic and measure how quickly it serves again from its last configuration.
Scenario.
The edge proxy was killed without warning while an application was called through it.
Expected behaviour.
- Restart and serve again within 5 seconds.
- Serve from the last known configuration.
Measured recovery.
- Recovery time RTO
- 0.4 s
- Target 5 s · met
- Application health
- Pass
Customer impact.
- Application availability
- 0 errors in 796 requests
What happened.
- Edge proxy killed.
- Serving again.
What we learned.
- The edge served again in 0.4 seconds, with no failed requests observed.
What we changed.
- No change needed.
Scope of this drill.
An edge proxy failure on one machine.