On Saturday, March 24, 2018, A System Admin <
asysad@gmail.com> wrote:
Hourly:
REINDEX TABLE miq_queue
REINDEX TABLE miq_workers
REINDEX TABLE metrics_$(date -u +"%H" --date='+1 hours ' )
Weekly:
\timing
VACUUM full verbose vms ;
vacuum full verbose binary_blob_parts ;
While I don't really have positive advice I do think you need to continue reading and asking questions since neither scheduled reindex nor scheduled vacuum fulls should generally be a necessary component of a maintenance routine.
Either per table or overall, more frequent vacuuming should be setup first - frequent enough to either solve the supposed problem (I'm not sure you really have one...) or until the vacuuming itself causes measurable pain. Depending on usage patterns manual vacuum (not full) could be done after bulk updates or deletes instead of waiting for the autovaccuum daemon to kick in.
David J.