Re: is cfbot's apply aging intentional?

On Sun, Mar 7, 2021 at 3:21 AM Julien Rouhaud <rjuju123(at)gmail(dot)com> wrote:
> On Sat, Mar 06, 2021 at 03:00:46PM +0100, er(at)xs4all(dot)nl wrote:
> >
> > I was looking at the 'Catalog version access' patch, by Vik Fearing. I saw a succesful build by the cfbot but I could not build one here. Only then did I notice that the last apply of the patches by cfbot was on 3769e11 which is the 3rd march, some 10 commits ago.
> >
> > There have been no new patches; one of the patches does not apply anymore. But it's not reported on the cfbot page.
> >
> > Is that the way it's supposed to be? I would have thought there was a regular schedule (hourly? 3-hourly? daily?) when all patches were taken for re-apply, and re-build, so that when a patch stops applying/building/whatever it can be seen on the cfbot page.
> >
> > Maybe I'm just mistaken, and the cfbot is supposed to only rebuild when there is a new patch. That would be kind-of logical too, although I for one would prefer a more continuous building.
> >
> > Can you tell me what is the intention at the moment? Is this a cfbot bug -- or just me being inadequately informed? ;)
>
> The cfbot will periodically try to rebuild all open patches on the current (and
> next) commitfest, as the main goal is to quickly spot patches that have rotten.
> But it's running on free platforms so Thomas put some mechanisms to avoid
> consuming too many resources. Looking at the source it won't try to rebuild
> any patch more than once per hour, and will try to have all patch rebuilt every
> 2 days in a best effort, at least with default configuration. Maybe there's
> some less aggressive setting on the deployed instance. So this patch will
> probably be checked soon.

Right, it currently tries to reassemble each branch every ~3 days. It
actually has a target of doing it every 2 days but it can't reach that
currently because it also has a constraint of not allowing more than 3
builds at once (I could increase that, but I was trying to be
polite...). It has around ~250 entries to get through, and they can
take up to 20 minutes to test. That said, most of that time is
actually wasted doing stupid stuff, and tweak by tweak, the scripting
it uses for CI builds is getting more efficient so the frequency will
hopefully soon increase. cfbot really needs to steal a whole lot of
fresh CI improvements from Andres, who has recently developed a set of
optimised, small fast-start disk images compatible with Cirrus, that
have all the right packages pre-installed, plus lots of other
improvements (test parallelism, better log extraction, better core
analysis, ...). More on that soon, hopefully.

But ... hmm, there must be something else going wrong for Erik here.
Vik's "Catalog version access" patches apply and compile and test fine
for cfbot and for me locally, with both GNU patch (what cfbot uses) or
git am (what I just tested with locally).