Le jeudi 1 mai 2025, 18:01:13 heure d’été d’Europe centrale Jochen Sprickerhof
a écrit :
> * Bastien Roucaries <ro...@debian.org> [2025-05-01 17:40]:
> >Le jeudi 1 mai 2025, 17:26:34 heure d’été d’Europe centrale Jochen
> >Sprickerhof>
> >a écrit :
> >> mmdebstrap --variant=apt --arch=i386 --chrooted-customize-hook=bash
> >> unstable /dev/null
> >
> >This one does not fully work for core dump....
> >
> >Try on the container:
> >#include <stdlib.h> // C
> >
> >void main(void)
> >{
> >
> > abort();
> >
> >}
> >
> >Do not get my core file in container /var/lib/systemd/coredump/ but in host
> >/ var/lib/systemd/coredump/
>
> That's a limitation of systemd-coredump or the kernel, depending on how
> you see it. From a quick look the pattern in
> /proc/sys/kernel/core_pattern is not namespaces so mmdebstrap can't
> change it but the kernel supports adding user namespaces to it:
>
> https://www.kernel.org/doc/html/latest/admin-guide/sysctl/kernel.html#core-p
> attern
>
> So systemd-coredump could do that but it would probably need to know the
> mmdebstrap internals to du so. So I guess disabling systemd-coredump
> would make the most sense.I suppose or pass the dump when you know the mechanism I do not ask a perfect solution, but a workable and documented on the wiki solution. For me long term, we need to have a user tagged bug chroot-arch problem and solve it in mmbootstrap If we want to have partial arch, chroot only arch, we need to document this kind of problem. Bastien > > Cheers Jochen
signature.asc
Description: This is a digitally signed message part.
