> I'm sorry for not having read (and not reading) the other thread yet, > but what was the reason we couldn't store that oid in a column in the > pg_s_p_vacuum-view?
> Could you summarize the other solutions that were considered for this issue? Thanks for your feedback! The reason we cannot stick the oid in pg_s_p_vacuum is because it will not work for parallel vacuum as only the leader process has an entry in pg_s_p_vacuum. With a function the leader or worker pid can be passed in to the function and will return the indexrelid being processed. Regards, Sami