13 / 18 · Module 16 — Production Engineering and Observability · filtered: cheatsheet← prev⊞ allnext →
16.41Scenario: retry count spikes
That itself is an operational signal.
Even if:
final success rate = 99.9%
a sudden jump from:
0.1 retries/request
to:
2.5 retries/request
means a dependency is degrading.
Retries can temporarily mask a real incident.
Monitor them.