There is no simple way to start a network installation of Fedora from USB stick. I have found a blog post that makes possible to transform a boot.iso image into a bootable USB stick. But it still requires to download the 200+ MB large iso file, while in theory only initrd.img and vmlinuz files should be needed. So I have created mkbootimg.sh script that automates all required steps and has a bootable disk image as a result. It can be transferred to real USB disk by the dd command. There is no warranty for the script, it can eat your system :-)

Red Hat Czech Republic is organizing a Developer Conference 2009 for Fedora developers, administrators and users during September. I plan to organize a workshop focused on secondary architectures in Fedora - s390, ARM, SPARC, building packages in Koji, preparing kernels for ARM platforms, etc. So if you are interested in participating in such event, let me know or just come in.

I was upgrading my workstation yesterday, only the case and harddisks remained the same. Almost everything went well, even new init ramdisk wasn't required, but I was stuck on networking for a while. And the problem was called "eth0_rename". After querying google it was clear that this is the result of "network device persistency" feature and the cause is different MAC address assigned to the new eth0 device. Solution was easy - remove the old records from
/etc/udev/rules.d/70-persistent-net.rules

The conservative side of Fedora called the Server SIG has been announced on the fedora-devel mailing list.

---
There were many discussions in the past few days and weeks about the orientation Fedora currently has. It is a fact that currently Fedora is primarily desktop oriented.

We agree that Desktop is important part of the system, it is highly visible to the public and large number of Fedora users. But we also see a large number of Fedora and CentOS users and RHEL customers with very specific needs and demands. We can not omit the server fundamentals that later create a successful enterprise product and in our opinion a formal entity must exist to coordinate these efforts.

That's why we started work on establishing the Fedora Server SIG. The draft is available at https://fedoraproject.org/wiki/DanHorak/ServerSIG
Any constructive ideas are welcome :-)

I am using stable (and quite old :-) ) Fedora 8 for doing my development work, but sometimes it is useful to run an application on Rawhide, for example during Package Reviews. And this task can be successfully accomplished using mock even for GUI applications.

Here are the required steps, first run in your host:
xhost +localhost (GUI only) + enable network connections to your X server
mock -r fedora-devel-x86_64 init
mock -r fedora-devel-x86_64 install app_you_want_to_run
cp -a /usr/share/X11/fonts /var/lib/mock/fedora-devel-x86_64/root/usr/share/X11 (GUI only)
mock -r fedora-devel-x86_64 shell

And then run in mock shell:
export DISPLAY=localhost:0.0 (GUI only)
app_you_want_to_run

A new window should appear on your desktop, when you are running a GUI application.

When I was working as a system/network admin, I used to have miscellaneous network configurations stored in my /etc/sysconfig/network-scripts/ifcfg-* files. There was a default DHCP config and some special setups with static IP addresses and gateways, all bound to the eth0 interface. It was in early Fedora 8 days and such setup was not compatible with NetworkManager I was using mainly for the easy setup of wifi networking. So I have wanted to know whether there was an improvement. Now I have used recent Fedora Rawhide, created ifcfg-test as a copy of ifcfg-eth0, but with staticly defined IP address/netmask and restarted NetworkManager and checked the applet. And voila - there were "System eth0" and "System test" items now and I could easily switch between them.