While investigating some 5xx issue in a cluster found the following stats for the cluster that has problems.

cluster.test_cluster.upstream_cx_none_healthy 10
cluster.test_cluster.lb_healthy_panic 10
cluster.test_cluster.membership_total : 3
cluster.test_cluster.membership_healthy: 3

While membership metrics show all the members as healthy, the cx_none_healthy and lb_healthy_panic incrementing seems to be inconsistent. Is this a bug or is there a scenario where this can happen?