Debugging why an observed query is refetched

[lwasyl]

In UI tests we're enqueueing responses for the network requests that we expect. However, sometimes Apollo makes additional requests that we don't expect (e.g. because a query doesn't fetch ID). What's the best way to monitor why Apollo is considering a query stale and in need for a network request to refetch it? I'm looking for something where I can e.g. log that a certain query is being observed and Apollo will now make a request for it, because some specific event happened/cache keys got modified