Recently I rebooted one of my Debian Xen servers and suddenly all the Ethernet devices which used to be eth0 in the domU’s became eth1.
vif = [ '', 'bridge=xenbr1' ]
I used to have the above as the interface definition and for domU’s that had only a single interface that worked well (if there is only one interface then it should be eth0). However in a recent etch update this changed, so I had to use ifrename as documented in my previous blog post. It’s annoying when things break because a reasonable assumption which previously worked suddenly stops working.
Even if the bug in question (if it is regarded as a bug) is fixed I’ll keep using ifrename, it doesn’t do any harm.
Update: I have changed my Xen configuration to use fixed MAC addresses which seems to be a better solution than using ifrename. See the Wikipedia page about MAC addresses for information on how to choose them. I’m currently using manually assigned MAC addresses from the range 00:16:3e (which is assigned to Xen).