high frequency failures of branch scanning is not alerted to ops

[Martin Pool]

Per today's canonical-launchpad thread "package-importer restart", it seems we don't have anything that alerts us if the branch scanner suddenly starts failing to scan many branches, as happened in bug 715000. We should expect some amount of failure through environmental errors but a sudden spike probably indicates something we should investigate.

[Robert Collins (lifeless)]

I'm going to drop this to low; while its a value report, we've been very happy with the overall reporting in place for this subsystem so far, and I doubt we'd get to this within 6 months.

[Martin Pool (mbp)]

> Generally we wait for a user to tell us if a given branch isn't
> scanning; we expect users to break branches from time to time and
> focus on having the scanner run and attempt to scan reliably.

That's true. However, if thousands of branches are failing, something's probably wrong beyond just normal background radiation. I agree with putting it to Low.