Question

A legacy Java application takes about 90 seconds to initialize before it can serve any traffic. You configured a livenessProbe with an httpGet on port 8080, initialDelaySeconds of 10, periodSeconds of 10, and failureThreshold of 3. The pod repeatedly enters CrashLoopBackOff during startup, but the application is otherwise healthy once running. Which change best resolves the restarts without weakening ongoing liveness detection?

Accepted Answer

Add a startupProbe with the same httpGet, periodSeconds 10, and failureThreshold 30, so the liveness probe does not begin until startup succeeds.

Answer

Increase the livenessProbe initialDelaySeconds to 120 to cover the initialization time.

Answer

Remove the livenessProbe entirely so the application is never restarted.

Answer

Set the livenessProbe failureThreshold to 60 so it tolerates the slow startup.