Comment 4 for bug 80990

The current heuristic used to guesstimate a mirror's freshness is not good for pockets which are not updated often (e.g. Backports, Proposed), so it often leaves these pockets on a days/weeks behind state. This means lots of mirrors will be disabled if we implement what's discussed on this bug, so we're not doing it until we sort out the other issue.

In order to develop a better heuristic we'd have to perform lots of tests and try several different approaches (which is going to take quite some time), so another option would be to skip all non-essential pockets (release, updates, security, ...) when probing. We may also want to skip the non-essential components (multiverse?).