Postgres Pro
Downloads
Home   >   mailing lists

Re: [Feature Request] INSERT FROZEN to Optimize Large Cold Data Imports and Migrations - Mailing list pgsql-hackers

From wenhui qiu
Subject Re: [Feature Request] INSERT FROZEN to Optimize Large Cold Data Imports and Migrations
Date
Msg-id CAGjGUA+entdtMAA5ShTe=_cXozTPj-VJtVdoP0S8T+kUp_ZGXw@mail.gmail.com
Whole thread Raw
In response to [Feature Request] INSERT FROZEN to Optimize Large Cold Data Imports and Migrations  (Sébastien <bokanist@gmail.com>)
List pgsql-hackers
Tree view
HI 
> Ok you mean that xid64 will remove the need for freezing... it's a way to see things.
When xid64 is implemented, there will be no need to trigger vacuum_failsafe_age,it  has a long enough time to vacuum freeze, it will have less of an impact on performance，I think that problem  may be due to  trigger the vacuum_failsafe_age


Thanks



On Wed, 19 Feb 2025 at 01:13, Sébastien <bokanist@gmail.com> wrote:
Ok you mean that xid64 will remove the need for freezing... it's a way to see things.

Le mar. 18 févr. 2025 à 15:57, Greg Sabino Mullane <htamfids@gmail.com> a écrit :
On Tue, Feb 18, 2025 at 9:17 AM Sébastien <bokanist@gmail.com> wrote:
Sorry it won't work. It just delays the problem. But still the freeze procedure must rewrite all pages.

Actually, a 64-bit transaction ID allows for quite a "delay" - like hundreds of millions of years at your current rate. :) 

(Yes, there are other reasons to vacuum, and other limits and problems would arise. You'd have 99 problems, but a vacuum freeze ain't one.)

Cheers,
Greg

--
Enterprise Postgres Software Products & Tech Support



--
Sébastien Caunes
+33 6 7 229 229 7

pgsql-hackers by date:

Previous
From: Richard Guo
Date:
Subject: Re: Unsafe access BufferDescriptors array in BufferGetLSNAtomic()
Next
From: Andy Fan
Date:
Subject: Re: Why does exec_simple_query requires 2 snapshots