MIME-Version: 1.0
References: <3ydjipcr7kbss57nvi67noplncqhesl5eyb6wgol4ccjxynspv@yatlykpribmm>
In-Reply-To: <3ydjipcr7kbss57nvi67noplncqhesl5eyb6wgol4ccjxynspv@yatlykpribmm>
From: Robert Haas <robertmhaas@gmail.com>
Date: Fri, 17 Apr 2026 14:50:53 -0400
Message-ID: 
 <CA+TgmoZJzNQ_ATdOi+WUZwc=SO5GjGKBbEBrMyLCmDZpkJ1D0Q@mail.gmail.com>
Subject: Re: Heads Up: cirrus-ci is shutting down June 1st
To: Andres Freund <andres@anarazel.de>
Cc: pgsql-hackers@postgresql.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Archived-At: 
 <https://www.postgresql.org/message-id/CA%2BTgmoZJzNQ_ATdOi%2BWUZwc%3DSO5GjGKBbEBrMyLCmDZpkJ1D0Q%40mail.gmail.com>
Precedence: bulk

On Thu, Apr 9, 2026 at 4:55=E2=80=AFPM Andres Freund <andres@anarazel.de> w=
rote:
> I'd be interested in feedback about how high folks value different aspect=
s:
>
> 1) CI software can be self hosted
> 2) CI software is open source
> 3) CI runs quickly
> 4) CI tests as many operating systems as possible
> 5) CI can be enabled on one's own repositories
> 6) There need to be free credits for running at least some CI on one's ow=
n
>    repository
> 7) Provide CI compute for "well known contributors" for free in their own
>    repositories

I think we need most of these things. CI has become an indispensable
development tool for most of us at this point. If (1) and (2) then (6)
and (7) are less necessary, and conversely. (5) seems pretty critical;
as long as I took to get CI set up on my own repo, I now use it
extensively. (4) is less critical: we could probably live with just
Linux and Windows in a pinch; adding MacOS and/or *BSD would be nicer.

> For some context about how much CI we have been running, here's the daily
> average for cfbot and postgres/postgres CI:
>
> - 1464 core hours (full cores, not SMT), all CI jobs use 4 cores
> - 396 core hours of which were windows (visible due to the licensing cost=
)
> - 40 GB of artifacts
> - 83 GB of artifacts downloaded externally
> - doesn't include macos, which I can't track as easily, due to being self
>   hosted runners, rather than running on GCP, which provided the above nu=
mbers

I wonder if we should be looking to add more heuristics to the system
to try to reduce these numbers. For example, just browsing through the
cfbot queue, I found this:

heapam_tuple_complete_speculative : remove unnecessary tuple fetch

https://cirrus-ci.com/github/postgresql-cfbot/postgresql/cf%2F6613

This patch removes six lines of code and adds none. There are four
messages on the thread. We've done 14 complete CI runs. That might be
an extreme example, but I just don't know if repeatedly running CI on
small patches that aren't being actively updated is really what we
want to be doing.

--=20
Robert Haas
EDB: http://www.enterprisedb.com