When we copy liveness to readiness, liveness should have increased failure threshold

[mortenlj]

When a user only specifies a liveness check, our current behavior is to copy it to the readiness check. Some blogs advocate that in the case where your liveness and readiness checks are the same, the liveness check should have a higher failure threshold.

This will allow your app to be considered unready and taken out of rotation, while still giving it some more tries to get ready before it is killed. This would help in cases where high load from the start makes it harder to come up.