Horton

Re: Cirrus-ci is lowering free CI cycles - what to do with cfbot, etc?

Andres Freund <andres@anarazel.de>

View thread View on postgr.es
From: Andres Freund <andres@anarazel.de>
To: Robert Treat <rob@xzilla.net>, sysadmins@lists.postgresql.org, Noah Misch <noah@leadboat.com>
Cc: Heikki Linnakangas <hlinnaka@iki.fi>, pgsql-hackers@postgresql.org, Thomas Munro <thomas.munro@gmail.com>, Justin Pryzby <pryzby@telsasoft.com>, Joe Conway <mail@joeconway.com>, "Jonathan S. Katz" <jkatz@postgresql.org>
Date: 2023-08-09T02:59:55Z
Lists: pgsql-hackers 
Hi,

On 2023-08-08 22:29:50 -0400, Robert Treat wrote:
> In case it's helpful, from an SPI oriented perspective, $7K/month is
> probably an order of magnitude more than what we can sustain, so I
> don't see a way to make that work without some kind of additional
> magic that includes other non-profits and/or commercial companies
> changing donation habits between now and September.

Yea, I think that'd make no sense, even if we could afford it. I think the
patches I've written should drop it to 1/2 already. Thomas added some
throttling to push it down further.


> Purchasing a couple of mac-mini's (and/or similar gear) would be near
> trivial though, just a matter of figuring out where/how to host it
> (but I think infra can chime in on that if that's what get's decided).

Cool. Because of the limitation of running two VMs at a time on macos and the
comparatively low cost of mac minis, it seems they beat alternative models by
a fair bit.

Pginfra/sysadmin: ^


Based on being off by an order of magnitude, as you mention earlier, it seems
that:

1) reducing test runtime etc, as already in progress
2) getting 2 mac minis as runners
3) using ~350 USD / mo in GCP costs for windows, linux, freebsd (*)

Would be viable for a month or three? I hope we can get some cloud providers
to chip in for 3), but I'd like to have something in place that doesn't depend
on that.

Given the cost of macos VMs at AWS, the only of the big cloud providers to
have macos instances, I think we'd burn pointlessly quick through credits if
we used VMs for that.

(*) I think we should be able to get below that, but ...


> The other likely option would be to seek out cloud credits from one of
> the big three (or others); Amazon has continually said they would be
> happy to donate more credits to us if we had a use, and I think some
> of the other hosting providers have said similarly at times; so we'd
> need to ask and hope it's not too bureaucratic.

Yep.

I tried to start that progress within microsoft, fwiw.  Perhaps Joe and
Jonathan know how to start within AWS?  And perhaps Noah inside GCP?

It'd be the least work to get it up and running in GCP, as it's already
running there, but should be quite doable at the others as well.

Greetings,

Andres Freund

Commits

  1. Use snprintf instead of sprintf in pg_regress.

  2. Speed up pg_regress server readiness testing.

  3. ci: Make compute resources for CI configurable

  4. ci: Prepare to make compute resources for CI configurable

  5. ci: Use VMs for SanityCheck and CompilerWarnings

  6. ci: Move execution method of tasks into yaml templates

  7. ci: Don't specify amount of memory

  8. ci: macos: Remove use of -Dsegsize_blocks=6

  9. ci: macos: Remove use of -DRANDOMIZE_ALLOCATED_MEMORY