On Thu, Jun 8, 2017 at 10:25 PM, Andres Freund <andres@anarazel.de> wrote:
It does so when truncating the relation, to remove free space at the end of the relation. If there's a lot of free space, that can take a while. This has been much improved in releases since 8.4.
Thank you for this information. Is it correct to assume that the VACUUM would NOT release its AccessExclusive lock during a vacuum_cost_delay sleep? Other than reading the PostgreSQL manual regarding Routine Vacuuming and the SQL reference for VACUUM, are there other documents or books you could recommend that would cover VACUUM internals (other than attempting to read source :-) )?