Re: rv-manda-01 might have hardware issues?
Hi,
On 2023-08-03 10:49, Adrian Bunk wrote:
> On Wed, Aug 02, 2023 at 07:08:18PM +0200, Manuel A. Fernandez Montecelo wrote:
> > Hi,
> >
> > On Wed, 2 Aug 2023 at 17:32, Adrian Bunk <bunk@debian.org> wrote:
> > > [...]
> > > This happens only on rv-manda-01, and my guess would be that this might
> > > be a hardware problem (e.g. a nonworking fan).
> > >
> > > Much worse than the build failures is that I wouldn't 100% trust that
> > > the contents of successful builds is not corrupted.
> >
> > Ack, will stop it for the moment, thanks.
>
> rv-manda-01 is running again, 2 questions:
>
> 1. Is rv-manda-01 now stable?
>
> The problem seems to have gone so far, has anything be fixed
> (or was it rebooted and that was sufficient)?
It got rebooted to a new kernel, but that was unrelated.
> 2. Rebuild of potentially broken packages.
>
> I would like to do "Rebuild on non-flaky buildd" binNMUs of the
> 984 packages that were built on rv-manda-01 before my previous
> email, because these packages cannot be trusted.[1]
As said in my previous mail, I don't think there is any hardware issue
with rv-manda-01. Or if there is, it's affecting all buildds. Therefore
please do not schedule those binNMUs.
> Example:
>
> iraf-mscred and iraf-fitsutil do FTBFS with weird errors on several
> buildds, after looking at the iraf-fitsutil error my first suspect
> would be that cfitsio was built on rv-manda-01.
I have been able to reproduce the issue locally, and it happens with
both cfitsio from the debian archive and the debian-ports archive.
Please note that the default gcc is now gcc-13 as opposed to gcc-12 for
most packages that were built on debian-ports archive. And it's not
unlikely that gcc-13 has issues, for instance I found that one affecting
glibc:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110066
> cfitsio (or some other package) having been built on rv-manda-01 might
> or might not be the root cause here, the problem is that we cannot rule
> it out and people might waste time debugging if it is.
>
> Opinions/Objections regarding me scheduling the 984 binNMUs?
Please refrain fro scheduling them until we see things more clear.
Cheers
Aurelien
--
Aurelien Jarno GPG: 4096R/1DDD8C9B
aurelien@aurel32.net http://aurel32.net
Reply to: