Nikola <XLPizza(at)gmail(dot)com> writes:
> Currently, I truncate the destination table, drop the indexes,
> execute:
> "INSERT INTO destination SELECT * FROM source_partition"
> and recreate indexes.
> This loads about 66.5 million rows into the destination table in about
> 1 hour. Recreating of indexes (4 of them) takes additional 2 hours. Is
> there a more efficient way to perform this procedure?
Do you have maintenance_work_mem set to a suitable value for building
the indexes?
regards, tom lane