| From: | Sami Imseih <samimseih(at)gmail(dot)com> |
|---|---|
| To: | Melanie Plageman <melanieplageman(at)gmail(dot)com> |
| Cc: | Konstantin Knizhnik <knizhnik(at)garret(dot)ru>, Aleksander Alekseev <aleksander(at)timescale(dot)com>, PostgreSQL Hackers <pgsql-hackers(at)postgresql(dot)org> |
| Subject: | Re: New criteria for autovacuum |
| Date: | 2025-04-07 02:18:21 |
| Message-ID: | CAA5RZ0sj5DHJASgKbyQFsA6msEb1JfRcWfr7=ozOZ-bG2AmPnw@mail.gmail.com |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
> I wasn't thinking about adding a new VM setting functionality to index
> only scan in particular. heapam_index_fetch_tuple() already calls
> heap_page_prune_opt() which will do pruning under certain conditions.
> I was thinking that we start updating the VM after pruning in the
> on-access case too (not just when pruning is invoked by vacuum).
I think this will need to be careful when the workload is constantly updating
and reading the same pages, as we may end up with a continuous cycle of
updates clearing the visibility-map and selects setting the visibility-map.
this may degrade the select workloads, maybe?
--
Sami Imseih
Amazon Web Services (AWS)
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Zhijie Hou (Fujitsu) | 2025-04-07 02:18:58 | RE: pg_recvlogical cannot create slots with failover=true |
| Previous Message | Anton A. Melnikov | 2025-04-07 01:55:45 | Re: Use XLOG_CONTROL_FILE macro everywhere? |