Breadcrumbs on the compute attributes detail page are tricky. Switcher lists compute profiles, but in fact the page url requires compute attribute id. As a result, the switcher links use compute profile ids in links to compute attributes. The links then lead to a seemingly random compute attributes (often to non-existent ones).

The eract switcher component always expects ids to be in 'id' attribute. One possible solution could be adding 'idField' attribute (we already have 'nameFiled') and enhance the compute profiles api with 'compute_attribute_id'. Alternatively we could use compute attributes api and add 'compute_profile_name'.