Re: Minimum tuple threshold to decide last pass of VACUUM

On Mon, Aug 3, 2015 at 4:13 AM, Simon Riggs <simon@2ndquadrant.com> wrote:
> * For normal VACUUMs we should scan indexes only if (num_dead_tuples * 20) >
> (blocks to be scanned in any one index), which allows some index bloat but
> not much

I think this kind of heuristic is good, but I think we should expose a
setting for it.  There's no way for us to know without testing whether
the right value for that multiplier is 2 or 20 or 200 or 2000, and if
we don't make it easy to tweak, we'll never find out.  It may even be
workload-dependent.

-- 
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company