| From: | Maxim Orlov <orlovmg(at)gmail(dot)com> |
|---|---|
| To: | Bharath Rupireddy <bharath(dot)rupireddyforpostgres(at)gmail(dot)com> |
| Cc: | Shinya Kato <shinya11(dot)kato(at)gmail(dot)com>, Scott Ray <scott(at)scottray(dot)io>, wenhui qiu <qiuwenhuifx(at)gmail(dot)com>, Sami Imseih <samimseih(at)gmail(dot)com>, Japin Li <japinli(at)hotmail(dot)com>, PostgreSQL Hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org> |
| Subject: | Re: Report oldest xmin source when autovacuum cannot remove tuples |
| Date: | 2026-11-18 15:03:00 |
| Message-ID: | CACG=ezb-_HQk7Z0MaBGek-trz6FLHTQxoVWFTsvvSjXUA5wk1w@mail.gmail.com |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
On Thu, 1 Oct 2026 at 15:21, Shinya Kato <shinya11(dot)kato(at)gmail(dot)com> wrote:
>
> v10 addresses all three.
Thank you for working!
I frequently work with clusters that have a large number of connections.
For this reason, I would like to raise a concern, perhaps small, but
significant to me. The XidHorizonBlocker struct contains a
name[NAMEDATALEN] field; when multiplied by the number of connections,
this can result in a substantial. All just so that, later on, we can
identify the "best" entry within that array.
But what if we try to find the "best" one while iterating through the loop?
I think that would be more optimal. Looks like this topic has already been
discussed before.
On Tue, 17 Nov 2026 at 01:05, Bharath Rupireddy <
bharath(dot)rupireddyforpostgres(at)gmail(dot)com> wrote:
>
> Thanks for the updated patch. Sorry for the delay. I have been
> thinking about how useful reporting the per-vacuum blocker will be
> with the pg_xmin_horizon view proposed in the other thread. Why can't
> users find the blockers with the view and the removable cutoff that is
> already reported in the verbose log output?
>
>
That's interesting. BTW, I do not see the pg_xmin_horizon patch CF entry.
I like the idea of being able to know what is blocking VACUUM, and the
proposed patch caught my interest. My concern, however, is also in the
need to iterate through the proc array multiple times. In my case, and I
have to deal with a very large number of connections, so that would
likely become an issue. OK, maybe not an issue, but slow things down
a bit.
As far as descusson goes, it looks like Fujii gives his approval on one
extra pgproc scan, does he? It would be great if he could also clarify
his position on the matter, if possible, especially in light of the
introduced pg_xmin_horizon patch.
--
Best regards,
Maxim Orlov.
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Heikki Linnakangas | 2026-11-18 15:16:40 | Re: Implicit shell type creation, little fixes |
| Previous Message | cca5507 | 2026-11-18 14:56:45 | Avoid copying MinimalTuple twice in ExecHashSkewTableInsert() |