FreeBSD CI Weekly Report 2019-03-24

[Kyle Evans]

On Tue, Mar 26, 2019 at 9:05 AM Enji Cooper <yaneurabeya at gmail.com> wrote:
>
>
> > On Mar 26, 2019, at 04:51, Kyle Evans <kevans at freebsd.org> wrote:
>
> ...
>
> > The lib.libc.regex.exhaust_test.regcomp_too_big failure should be
> > fixed by r345516 pulling in an rlimit bump from upstream. They didn't
> > adjust the test metadata though -- it previously reflected a memory
> > requirement of 64M, which matched the rlimit imposed. I would expect
> > that needs increased if we're exhausting 64M like we were on some
> > systems, but I'm unsure if we should just bump that sucker to 256M or
> > try to find an intermediate that's sufficient. I suspect the 256M bump
> > wasn't a measurement of usage.
> >
> >> [... snip ...]
>
> Memory serves me correctly, it used to time out with this value before, which is why pho@ and I lowered the limit to 64MB (I think we added the setrlimit call).
>
> I’ll have to go back and refresh my memory, since this was 3-4 years ago.
>

Looking back at the NetBSD history, I do see that you upstreamed the
64M limit and dropped the requirements from 120M to 64M in the
metadata. =-( Apologies for not researching that more closely- I had
only seen that they upped the setrlimit ~two weeks ago and assumed it
was safe at this point.