Re: Fun fact about autovacuum and orphan temp tables

From: Michael Paquier <michael(dot)paquier(at)gmail(dot)com>
To: Haribabu Kommi <kommi(dot)haribabu(at)gmail(dot)com>
Cc: Dilip Kumar <dilipbalaut(at)gmail(dot)com>, PostgreSQL mailing lists <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: Fun fact about autovacuum and orphan temp tables
Date: 2016-11-17 04:14:01
Message-ID: CAB7nPqTkoJ5QOkU8ch_-djDFYhn26_zjQZDkUAOrNT1R5ACepg@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Wed, Nov 16, 2016 at 5:24 PM, Haribabu Kommi
<kommi(dot)haribabu(at)gmail(dot)com> wrote:
> This is a gentle reminder.
>
> you assigned as reviewer to the current patch in the 11-2016 commitfest.
> But you haven't shared your review yet. Can you please try to share your
> views
> about the patch. This will help us in smoother operation of commitfest.

Thanks for the reminder.

> Michael had sent an updated patch based on some discussion.
> Please Ignore if you already shared your review.

Hm. Thinking about that again, having a GUC to control if orphaned
temp tables in autovacuum is an overkill (who is going to go into this
level of tuning!?) and that we had better do something more aggressive
as there have been cases of users complaining about dangling temp
tables. I suspect the use case where people would like to have a look
at orphaned temp tables after a backend crash is not that wide, at
least a method would be to disable autovacuum after a crash if such a
monitoring is necessary. Tom has already stated upthread that the
patch to remove wildly locks is not acceptable, and he's clearly
right.

So the best move would be really to make the removal of orphaned temp
tables more aggressive, and not bother about having a GUC to control
that. The patch sent in
https://www.postgresql.org/message-id/CAB7nPqSRYwaz1i12mPkH06_roO3ifgCgR88_aeX1OEg2r4OaNw@mail.gmail.com
does so, so I am marking the CF entry as ready for committer for this
patch to attract some committer's attention on the matter.
--
Michael

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Michael Paquier 2016-11-17 04:18:04 Re: Re: [sqlsmith] FailedAssertion("!(XLogCtl->Insert.exclusiveBackup)", File: "xlog.c", Line: 10200)
Previous Message Michael Paquier 2016-11-17 04:04:36 Re: Password identifiers, protocol aging and SCRAM protocol