Bridged networking in virt-manager

I have an ongoing inability to commit to a single virtualization platform because I’m always looking for one that has everything I need, is free and open source, and runs well on my dinosaur hardware.  Maybe I’ll post another time about my experiences with VMWare, Qemu, VirtualBox, and Xen.

Anyway, I just got a x86_64 KVM capable box from a co-worker that a is a few years old, but it is a few years newer than my existing hardware   I did a fresh install of Fedora 9, installed the Virtualization group, and set off to do a network install using virt-manager.  That was when I discovered that the shared physical device interface option was grayed out and couldn’t be used.

I went hunting around on Google for the solution, but didn’t find anything obvious right away.  In the end I found this excellent page: http://wiki.libvirt.org/page/Networking–oddly enough linked to in the comments of some instructions that didn’t work.  In all likelihood you do not need to do the  # cat > ifcfg-eth0 <<EOF step as you probably already have a working network card.  I did not add the iptables (firewall) rules either.

Now to get a rawhide tree that installs.  In the meantime I’m also experimenting with this very interest project: oVirt.

Fedora 10 Schedule Update

Last week the Fedora Engineering Steering Committee (FESCo) ratified the updated schedule proposed by the Release Engineering team.  This resulted in feature freeze for Fedora moving to 2008-09-09 and GA to 2008-11-18.  This three week change to the schedule was to accommodate the recent infrastructure outages.

Updated general milestones are here: https://fedoraproject.org/wiki/Releases/10/Schedule

For all the project management junkies out there, detailed tasks and durations here: http://poelstra.fedorapeople.org/schedules/f-10/f-10-all-tasks.html .  You can optionally filter tasks by team using the links at the top of each page.

I think it is great that we were willing to take such a big slip to the original schedule now.  In the past we’ve waited until the last minute to see if we could absorb slips at various milestones with the regular result of not being able to do so. With a 6 month release cycle the Fedora schedules are already compressed with very little cushion or room for things to go wrong built in.

From a project management perspective slipping releases at the last possible date is generally a bad idea because it doesn’t give other groups who depend on other tasks being completed first to respond or adjust expectations.  Examples in Fedora include artwork, documentation, infrastructure, and marketing.

UPDATE (2008-10-13): See post about upated schedule and Fedora 10 feature list here