On Sat, 2016-07-30 at 18:12 -0600, Chris Murphy wrote:
On Sat, Jul 30, 2016 at 5:48 PM, Chris Murphy
<lists(a)colorremedies.com> wrote:
>
> On Sat, Jul 30, 2016 at 4:41 PM, Chris Murphy <lists(a)colorremedies.com>
wrote:
> >
> > Hi,
> >
> > Using Fedora-Workstation-netinst-x86_64-25-20160730.n.0.iso in a
> > Fedora 24 virt-manager VM, I'm getting "failed to setup base
> > repository" from the installer. In a shell, ping fails to resolve, and
> > also says the network is unreachable if I ping an IP directly.
> > NetworkManager says the link is ready and up; ip -s a shows nothing
> > unusual, there is an IP assigned and it's sane.
> >
> > Any ideas?
>
> Same VM with F24 netinstall, there's no problem. And
> Fedora-Xfce-Live-x86_64-25-20160730.n.0.iso likewise has no problem.
> Huh.
OpenQA shows something similar
Fedora-Everything-netinst-x86_64-25-20160730.n.0.iso works in openqa.
https://openqa.fedoraproject.org/tests/26385
Fedora-Server-netinst-x86_64-25-20160730.n.0.iso also works in openqa
up until a different problem happens, not related to software
selection.
Fedora-Workstation-netinst-x86_64-25-20160730.n.0.iso in openqa gets
stuck on "error checking software selection" so it's a workstation
netinstall specific problem.
https://bugzilla.redhat.com/show_bug.cgi?id=1358416
it should be fixed with tomorrow's compose. The Workstation netinst
failure was not this bug, though; 'Error checking software selection'
is entirely different from 'Failed to setup base repository', it's a
much later failure, and is usually caused by dependency issues. That
one was most likely the libreoffice bug.
It's actually kind of odd that most of the openQA network install tests
*didn't* hit this bug, I think it has something to do with details of
the network configuration and the resulting NetworkManager composite
dhclient config file. At least one openQA test *is* affected, though,
and that's install_kickstart_nfs:
https://openqa.fedoraproject.org/tests/27314
which uses a slightly different network configuration (though oddly
enough some other tests which use the *same* network configuration
still work, so in summary...I don't know, what the hell).
Anyhoo, the underlying cause in dracut should be fixed tomorrow, so
this mess should hopefully go away.