Re: emba.gnu.org overload

[Michael Albinus]

Mattias Engdegård <mattiase@acm.org> writes:

> Hello Michael,

Hi Mattias,

emba related discussions happen on <emacs-build-automation@gnu.org>.
I've Cc'ed it.

> It seems that emba.gnu.org suffers from severe overload which has only
> become worse with the introduction of extra builds for the emacs-29
> branch. Without more resources, demand must be reduced -- as it is, it
> has become near-useless from timeouts and delayed completion.

Yes.

> I spent some time speeding up one of the worst offenders,
> ucs-normalize-tests, quite a bit but it seems it wasn't enough. What
> about putting the full builds of master on ice for the time being,
> while emacs-29 is stabilised? Or should we cut more of the slow tests
> (maybe all of ucs-normalize-tests)?

Well, the scheduled pipelines on emba contain 10 jobs. I believe we
don't need all of them.

For example, four different jobs for native-compilation builds are run,
build-native-comp-speed{0,1,2} and test-native-comp-speed0. That was
fine for development of this feature, but perhaps not so important these
days. We might ask Andrea Corallo about.

Also the platform tests build-image-{filenotify-gio,gnustep} and
test-{filenotify-gio,gnustep} could be reduced. We don't need them in
both emacs-29 and master pipelines; for the time being we might disable
them in master.

For ucs-normalize-tests, I don't know anything about. It times out after
1 hour, although the log is short:

--8<---------------cut here---------------start------------->8---
Running 7 tests (2022-12-05 04:29:17+0000, selector `(not (or (tag :unstable) 
(tag :nativecomp)))')
   passed  1/7  ucs-normalize-part0 (0.174699 sec)
UCS Normalize Test Part1, rule 1... 
UCS Normalize Test Part1, rule 1...1% 
UCS Normalize Test Part1, rule 1...2% 
UCS Normalize Test Part1, rule 1...3% 
UCS Normalize Test Part1, rule 1...4% 
UCS Normalize Test Part1, rule 1...5% 
UCS Normalize Test Part1, rule 1...6% 
UCS Normalize Test Part1, rule 1...7% 
UCS Normalize Test Part1, rule 1...8% 
UCS Normalize Test Part1, rule 1...9% 
UCS Normalize Test Part1, rule 1...10% 
UCS Normalize Test Part1, rule 1...11% 
UCS Normalize Test Part1, rule 1...12% 
UCS Normalize Test Part1, rule 1...13% 
UCS Normalize Test Part1, rule 1...14% 
UCS Normalize Test Part1, rule 1...15% 
UCS Normalize Test Part1, rule 1...16% 
UCS Normalize Test Part1, rule 1...17% 
UCS Normalize Test Part1, rule 1...18% 
UCS Normalize Test Part1, rule 1...19% 
UCS Normalize Test Part1, rule 1...20% 
UCS Normalize Test Part1, rule 1...21% 
UCS Normalize Test Part1, rule 1...22% 
UCS Normalize Test Part1, rule 1...23% 
UCS Normalize Test Part1, rule 1...24% 
UCS Normalize Test Part1, rule 1...25% 
UCS Normalize Test Part1, rule 1...26% 
UCS Normalize Test Part1, rule 1...27% 
UCS Normalize Test Part1, rule 1...28% 
UCS Normalize Test Part1, rule 1...29% 
UCS Normalize Test Part1, rule 1...30% 
UCS Normalize Test Part1, rule 1...31% 
UCS Normalize Test Part1, rule 1...32% 
UCS Normalize Test Part1, rule 1...33% 
UCS Normalize Test Part1, rule 1...34% 
UCS Normalize Test Part1, rule 1...35% 
UCS Normalize Test Part1, rule 1...36% 
UCS Normalize Test Part1, rule 1...37% 
UCS Normalize Test Part1, rule 1...38% 
UCS Normalize Test Part1, rule 1...39% 
UCS Normalize Test Part1, rule 1...40% 
UCS Normalize Test Part1, rule 1...41% 
UCS Normalize Test Part1, rule 1...42% 
UCS Normalize Test Part1, rule 1...43% 
UCS Normalize Test Part1, rule 1...44% 
UCS Normalize Test Part1, rule 1...45% 
UCS Normalize Test Part1, rule 1...46% 
UCS Normalize Test Part1, rule 1...47% 
UCS Normalize Test Part1, rule 1...48% 
UCS Normalize Test Part1, rule 1...49% 
UCS Normalize Test Part1, rule 1...50% 
UCS Normalize Test Part1, rule 1...51% 
UCS Normalize Test Part1, rule 1...52% 
UCS Normalize Test Part1, rule 1...53% 
UCS Normalize Test Part1, rule 1...54% 
UCS Normalize Test Part1, rule 1...55% 
UCS Normalize Test Part1, rule 1...56% 
UCS Normalize Test Part1, rule 1...57% 
UCS Normalize Test Part1, rule 1...58% 
UCS Normalize Test Part1, rule 1...59% 
UCS Normalize Test Part1, rule 1...60% 
UCS Normalize Test Part1, rule 1...61% 
UCS Normalize Test Part1, rule 1...62% 
UCS Normalize Test Part1, rule 1...63% 
UCS Normalize Test Part1, rule 1...64% 
UCS Normalize Test Part1, rule 1...65% 
UCS Normalize Test Part1, rule 1...66% 
UCS Normalize Test Part1, rule 1...67% 
UCS Normalize Test Part1, rule 1...68% 
UCS Normalize Test Part1, rule 1...69% 
UCS Normalize Test Part1, rule 1...70% 
UCS Normalize Test Part1, rule 1...71% 
UCS Normalize Test Part1, rule 1...72% 
UCS Normalize Test Part1, rule 1...73% 
UCS Normalize Test Part1, rule 1...74% 
UCS Normalize Test Part1, rule 1...75% 
UCS Normalize Test Part1, rule 1...76% 
UCS Normalize Test Part1, rule 1...77% 
UCS Normalize Test Part1, rule 1...78% 
UCS Normalize Test Part1, rule 1...79% 
UCS Normalize Test Part1, rule 1...80% 
UCS Normalize Test Part1, rule 1...81% 
UCS Normalize Test Part1, rule 1...82% 
UCS Normalize Test Part1, rule 1...83% 
UCS Normalize Test Part1, rule 1...84% 
UCS Normalize Test Part1, rule 1...85% 
UCS Normalize Test Part1, rule 1...86% 
UCS Normalize Test Part1, rule 1...87% 
UCS Normalize Test Part1, rule 1...88% 
UCS Normalize Test Part1, rule 1...89% 
UCS Normalize Test Part1, rule 1...90% 
UCS Normalize Test Part1, rule 1...91% 
UCS Normalize Test Part1, rule 1...92% 
UCS Normalize Test Part1, rule 1...93% 
UCS Normalize Test Part1, rule 1...94% 
UCS Normalize Test Part1, rule 1...95% 
UCS Normalize Test Part1, rule 1...96% 
UCS Normalize Test Part1, rule 1...97% 
UCS Normalize Test Part1, rule 1...98% 
UCS Normalize Test Part1, rule 1...99% 
UCS Normalize Test Part1, rule 2... 
UCS Normalize Test Part1, rule 2...1% 
UCS Normalize Test Part1, rule 2...2% 
UCS Normalize Test Part1, rule 2...3% 
UCS Normalize Test Part1, rule 2...4% 
UCS Normalize Test Part1, rule 2...5% 
UCS Normalize Test Part1, rule 2...6% 
UCS Normalize Test Part1, rule 2...7% 
UCS Normalize Test Part1, rule 2...8% 
timeout: the monitored command dumped core
--8<---------------cut here---------------end--------------->8---

Doesn't this mean rather a bug we need to capture?

> Emacs really needs a reliable and useful CI server and emba.gnu.org is
> the closest we've got. Hydra was nice as well but Glenn apparently
> found it to be a thankless task to maintain it (can't blame him) and
> stopped bothering, so it's failed all builds for months for machine
> configuration reasons.

Well, for months I'm the only one who maintains emba. Also not a very
stable constellation ...

> Cheers,
>
> Mattias

Best regards, Michael.