Horton
Menu

Re: Control your disk usage in PG: Introduction to Disk Quota Extension

Xing Guo <higuoxing@gmail.com>

View thread
From: Xing Guo <higuoxing@gmail.com>
To: Stephen Frost <sfrost@snowman.net>
Cc: Daniel Gustafsson <daniel@yesql.se>, Haozhou Wang <hawang@pivotal.io>, David Steele <david@pgmasters.net>, Michael Paquier <michael@paquier.xyz>, Alvaro Herrera <alvherre@2ndquadrant.com>, Hubert Zhang <hzhang@pivotal.io>, Thomas Munro <thomas.munro@gmail.com>, Andres Freund <andres@anarazel.de>, Robert Haas <robertmhaas@gmail.com>, "pgsql-hackers@postgresql.org" <pgsql-hackers@postgresql.org>, Tomas Vondra <tomas.vondra@2ndquadrant.com>, Heikki Linnakangas <hlinnakangas@pivotal.io>
Date: 2024-02-27T01:41:45Z
Lists: pgsql-hackers 
On Tue, Feb 27, 2024 at 6:38 AM Stephen Frost <sfrost@snowman.net> wrote:
>
> Greetings,
>
> * Xing Guo (higuoxing@gmail.com) wrote:
> > Looks that many hackers are happy with the original patch except that
> > the original patch needs a simple rebase, though it has been 3 years.
>
> I'm not completely against the idea of adding these hooks, but I have to
> say that it's unfortunate to imagine having to use an extension for
> something like quotas as it's really something that would ideally be in
> core.
>
> > Shall we push forward this patch so that it can benefit extensions not
> > only diskquota?
>
> Would be great to hear about other use-cases for these hooks, which
> would also help us be comfortable that these are the right hooks to
> introduce with the correct arguments.  What are the other extensions
> that you're referring to here..?

Sorry, we don't have another concrete extension that utilizes the
smgrcreate / smgrtruncate / smgrdounlinkall hooks right now. But we
have a transparent data encryption extension that utilizes the
smgrread and smgrwrite hooks to mutate the read buffer and write
buffer in place to perform file level encryption and decryption. I
think smgrwrite / smgrread / smgrcreate / smgrtruncate and
smgrdounlinkall hooks fall into the same catagory and these hooks are
different from the recent proposed storage manager API hooks[1].
Probably it is worth starting a new thread and discussing them.

[1]: https://www.postgresql.org/message-id/flat/CAEze2WgMySu2suO_TLvFyGY3URa4mAx22WeoEicnK%3DPCNWEMrA%40mail.gmail.com

> Thanks,
>
> Stephen

Best Regards,
Xing