I have recently made som efforts to jumpstart Solaris machines (Sparc hardware, enterprise servers) from a linux portable (Running 2.2.16 at the time). Unfortunately, although the Linux tftp/bootparamd services are fine, the first time the sparc tries to access the NFS share of the boot image, it fails. Having run a snoop on the connection whilst trying this (both from the linux machine and also a sun workstation) I've managed to (I think) trace the problem down. The problem seems to be that the NFS packets are fragmented in reverse on the linux machine, and in the 'correct' order by the solaris workstation. The kernel that is used to bootstrap solaris 2.7 doesn't seem to be able to cope with this. Once the boot share has been handled, the linux machine can be used for all subsequent stages of the boot, it is simply that first share that does not work. Does anyone have any advice, or know how easy it would be to change the method of fragmentation? - : send the line "unsubscribe linux-net" in the body of a message to majordomo@vger.kernel.org
