The first step of k-medoids is that k-centroids/medoids are randomly picked from your dataset (X), how those centroids are recalculated when there are some clusters changing
