I'm trying to automate deployment of a server with a single ethernet interface in a way that easily permits vm guests and/or vtun clients to share the (layer-2, ethernet) connection of the host.
What i need is for the host to have its own IP address on the bridge interface (e.g br0), which enslaves the physical interface (eth0), and then allow guest or tunnel clients to automatically add their interfaces to br0 (e.g. via "brctl addif br0 <new_dev>").
I found a couple of places on the Web that show how this can be retrofitted to NetworkManager running on an already existing machine:
https://avi.alkalay.net/2014/01/fedora-20-virtualization-networkmanager-nati...
https://www.happyassassin.net/2014/07/23/bridged-networking-for-libvirt-with...
but what I'd really like is to figure out how to express this during server build via anaconda/kickstart.
Looking through http://fedoraproject.org/wiki/Anaconda/Kickstart the "--bond*" and "--team*" options look intrigueing, but details on specific options (e.g. "--bondopts=mode=<foo>") are hard to come by.
Would "--bondopts" and "--bondslaves" actually allow me to create a *bridge* rather than an *etherchannel* ? If so, any place to RTFM on the details ? If not, is there something else I could use ?
Ultimately, I can probably hack /etc/sysconfig/networ-scripts/ifcfg-eth0 and add /etc/sysconfig/network-scripts-ifcfg-br0 during %post, but I'd really rather not do that if there's a "canonical" way of getting the job done :)
Thanks much, --Gabriel
Hi,
we are adding bridge support to installer (--bridge* opts), but it will make it only to Fedora 22, so for now using the %post kickstart section would the way to configure the bridge automatically during installation.
Radek
On 11/10/2014 08:44 PM, Gabriel Somlo wrote:
I'm trying to automate deployment of a server with a single ethernet interface in a way that easily permits vm guests and/or vtun clients to share the (layer-2, ethernet) connection of the host.
What i need is for the host to have its own IP address on the bridge interface (e.g br0), which enslaves the physical interface (eth0), and then allow guest or tunnel clients to automatically add their interfaces to br0 (e.g. via "brctl addif br0 <new_dev>").
I found a couple of places on the Web that show how this can be retrofitted to NetworkManager running on an already existing machine:
https://avi.alkalay.net/2014/01/fedora-20-virtualization-networkmanager-nati...
https://www.happyassassin.net/2014/07/23/bridged-networking-for-libvirt-with...
but what I'd really like is to figure out how to express this during server build via anaconda/kickstart.
Looking through http://fedoraproject.org/wiki/Anaconda/Kickstart the "--bond*" and "--team*" options look intrigueing, but details on specific options (e.g. "--bondopts=mode=<foo>") are hard to come by.
Would "--bondopts" and "--bondslaves" actually allow me to create a *bridge* rather than an *etherchannel* ? If so, any place to RTFM on the details ? If not, is there something else I could use ?
Ultimately, I can probably hack /etc/sysconfig/networ-scripts/ifcfg-eth0 and add /etc/sysconfig/network-scripts-ifcfg-br0 during %post, but I'd really rather not do that if there's a "canonical" way of getting the job done :)
Thanks much, --Gabriel
Anaconda-devel-list mailing list Anaconda-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/anaconda-devel-list
anaconda-devel@lists.stg.fedoraproject.org