Hi,

Has anybody worked out how to do a network install from a node that is 
setup for jumbo frames?  E.g. a cluster head-node is set to use jumbo 
frames over internal network.  Client node is installed using anaconda + 
kickstart file with NFS used for retrieving distro from head-node.  The 
old solution that I've read is to switch the network back to 1500 byte 
frames for the install --- which is a problem if the rest of the cluster 
is in production.  A better work around is to setup an install host that 
is set for 1500 byte frames, though that requires another node...

This isn't a show-stopper for us right now as we have a second network 
that is setup for 1500 byte frames.

I guess you can probably set the MTU as a kernel option when booting the 
install kernel, but then you also have to convince anaconda to set jumbo 
frames when it restarts the network at the beginning of the install.

Thanks,
Tom