pg.ddx.io  pgbackrest/pgbackrest GitHub issues and pull requests  
help / color / mirror / Atom feed
From: dwsteele (@dwsteele) <noreply+dwsteele@github.com>
To: pgbackrest/pgbackrest <noreply+pgbackrest-pgbackrest@github.com>
Subject: Re: [pgbackrest/pgbackrest] PR #2887: DirectIO
Date: Wed, 04 Nov 2026 15:19:00 +0000
Message-ID: <gh-pgbackrest-pgbackrest-2887-c6634426981@github.com> (raw)
In-Reply-To: <gh-pgbackrest-pgbackrest-2887@github.com>
References: <gh-pgbackrest-pgbackrest-2887@github.com>

> I have applied the requested changes

This looks a lot more like what I was expecting, thank you!

> If O_DIRECT is slow we could add io_uring

I briefly looked at io_uring and was put off a bit by the complexity thinking we should stick with fadvise() but it might not look like such a big diff if DIO goes in.

> RWF_DONTCACHE is interesting, we could add this without flag and auto-discovery, but it would not work for any current RHEL versions.

Yeah, this would be best effort and fail-back to default. But it seems to be the best possible option for distros where it works, and those will only grow.

I won't be able to look at this immediately but the goal would be to get it into the January release as a beta feature, as long as there are no good reasons to keep it out.

view thread (20+ messages)  latest in thread

Message-ID: <gh-pgbackrest-pgbackrest-2887-c6634426981@github.com>
Permalink:  ../gh-pgbackrest-pgbackrest-2887-c6634426981@github.com/

 · 


This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox