As described in OSSM-3392, when creating a destination rule for an imported service, a single unhealthy pod can trigger a circuit breaker for the imported service, regardless of how many otherwise healthy pods may be available.

Expected Behaviour

A single unhealthy pod should not trigger a circuit breaker for the entire imported service. Instead, it should ideally only create a circuit breaker for the unhealthy pod and continue to send requests to other healthy pods of the imported service.

Failing this, it should be possible to configure retries on the remote cluster so that one unhealthy pod is less likely to trigger a circuit breaker.