On Tue, Dec 10, 2013 at 7:55 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> This doesn't make me happy. Aside from the sheer waste of cycles
> involved in re-analyzing the entire regression database, this
> test runs in parallel with half a dozen others, and it could cause
> plan instability in those. Of course, if it does, then most likely
> those tests have a hazard from autovacuum anyway. But this still
> looks to me like a poor bit of test design.
Agreed.
> Anyway, bottom line is that I think we need to institute, and
> back-patch, some consistent scheme for when to analyze the standard
> tables during the regression tests, so that we don't have hazards
> like this for tests that want to check what plan gets selected.
>
> Comments?
Everything you're saying sounds reasonable from here.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company