If two networks have their names swapped in the server data the client hits an endless retry loop trying to update either and failing because of unique constraint errors.
Normally we are able to fix this on the server side by pushing a noop update to the other object, so the client fetches it last to resolve the conflic, but since its a 1:1 name swap this does not work here.