On Mon, 22 Dec 2014 17:43:39 -0800, Adam Williamson
<adamwill@xxxxxxxxxxxxxxxxx> said:
Once again I noticed a new nightly test event was 'needed' while
testing relval changes! So the new testing nightly is 20141222.
I've tested the x86_64 network install image on real hardware for the
Server. It did work.
We use a 1998 vintage DHCP server that's been heavily modified. As a
consequence, I have to specify enough networking specifications so that
a network installer can see the, um, network as Fedora 21 and later
doesn't want to send the hardware id as the defaul any longer. There's
some discussion of this in this closed BZ
(https://bugzilla.redhat.com/show_bug.cgi?id=1142889).
In the past, I've been able to switch to one of the text consoles once
the first panel is up, provide enough DHCP config to satisfy my vintage
server, and then return. It appears that this most recent
nightly-to-be-tested tries to reach back home and refresh the repos
before / during the display of that first panel. And it crashes
horribly at that point (usual panel with a Report this bug button). I'll
find some way to get the network up in the crashed system and submit a
problem next week, but didn't want to leave this unremarked until then.
I can specify enough information as kernel parameters to avoid this, but
that also raised an interesting question: on a new piece of hardware, if
one has to specify networking on the kernel parameter line, how does one
know what the name of the (or at least a) network device is?
Regards,
Robert
--
devel mailing list
devel@xxxxxxxxxxxxxxxxxxxxxxx
https://admin.fedoraproject.org/mailman/listinfo/devel
Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
