Question 220
A data center experiences intermittent leaf-to-spine packet loss while BGP sessions reset slowly, and the team compares only tenant MAC routes for a diagnosis. Once the issue is resolved, which ongoing practice BEST prevents this type of failure?
Explanation
Correct Answer
A lasting control should detect the original error before release. Monitor physical health and underlay failure detection separately from overlay advertisements. That is important in this case because underlay path failure detection and route convergence have their own timers and signaling independent of the tenant EVPN service database. The scenario was: A data center experiences intermittent leaf-to-spine packet loss while BGP sessions reset slowly, and the team compares only tenant MAC routes for a diagnosis.
Incorrect Answers
EVPN control-plane MAC mobility handling updates the advertised endpoint location and withdraws or supersedes stale location information. Here, underlay path failure detection and route convergence have their own timers and signaling independent of the tenant EVPN service database.
Include controlled link-failure ECMP convergence tests in rack commissioning and uplink changes. That concerns ECMP resilience after a spine link failure, while the issue here is underlay failure isolation using BFD and route state. Correlate interface errors, supported BFD or routing-neighbor state and route withdrawal, then correct the underlay fault and validate recovery.
VXLAN adds outer transport headers, making the required underlay packet size greater than the original tenant Ethernet payload. However, the present prevention decision is grounded in: A data center experiences intermittent leaf-to-spine packet loss while BGP sessions reset slowly, and the team compares only tenant MAC routes for a diagnosis. For this issue, correlate interface errors, supported BFD or routing-neighbor state and route withdrawal, then correct the underlay fault and validate recovery.