> > I have a diskless workstation setup (which I had working fine) and am > having trouble incorporating 100Mbps NICs into the picture. > > I have 10Mbps NICs and 100Mbps NICs, but netboot and etherboot only > support the 10Mbps NICs I have. > > However there are ways around this - just have the workstation use the > 10Mbps NIC to get an IP, grab its kernel, mount root via NFS and begin > booting. While booting, it configures its 100Mbps NIC and remounts root > over the 100Mbps line.
Are you trying to get a second IP assigned when the 100mbps card comes up or are you using the same IP? The same IP won't work because dhcp grabs NIC MAC addresses, so you must use a different IP for each card. > The problem is that when trying to mount root via the 100Mbps line I keep > getting permission denied error messages. So it is failing at literally > the very last step. Check the network names and hosts and exports. > here are the relevant details: > > The network is setup such the 10Mbps and 100Mbps networks are entirely > sepparate. Physically, there are no connectiosn between the two, and it > is configured so that 192.168.0.* is 10Mbps and 192.168.1.* is 100Mbps. > > ---------------------------------------------------- > Server: (seele) > > ----------------------------------------------------- > /etc/hosts > > 127.0.0.1 seele localhost > 192.168.0.1 seele localhost > 192.168.1.1 seele localhost > 192.168.0.2 casper > 192.168.1.2 casper I would not suggest that you do this - having the same name in a hosts file for different IPs is a cause of confusion. I would put .0.1 as seele_10 and .1.1 as seele_100 and likewise with casper. > > ---------------------------------------------------- > /etc/network/interfaces > > auto lo > iface lo inet loopback > > # internet 100Mbit connection (top card) > auto eth2 > iface eth2 inet static > address 192.168.1.1 > netmask 255.255.255.0 > > # internet 10Mbit connection (middle card) > auto eth0 > iface eth0 inet static > address 192.168.0.1 > netmask 255.255.255.0 > > # external 10Mbit connection (bottom card) > auto eth1 > iface eth1 inet dhcp Um, this doesn't make sense - you have three cards of which two are internet? Do you not mean intranet? Also, if you have these cards assigned IPs then what is the dhcp doing on eth1? or is that for cable modem or equiv? > > ------------------------------------------------------ > /etc/bootptab > > default:ds=192.168.0.1:\ > sm=255.255.255.0:\ > gw=192.168.0.1:\ > ds=66.245.60.4 > > casper:ht=ethernet:\ > sa=192.168.0.1:\ > ha=0x52544c1d2fb1:\ > ip=192.168.0.2:\ > tc=default:\ > hd=/tftpboot/casper:\ > bf=kernel:\ > rp=/nfsroot/casper: > > ------------------------------------------------------ > /etc/exports > > /nfsroot/casper/ 192.168.0.2(rw,no_root_squash) > /nfsroot/casper/ 192.168.1.2(rw,no_root_squash) Are you sure that you're allowed multiple entries? Try simply putting a single entry like this: /nfsroot/casper 192.168.0.0/255.255.0.0(rw,no_root_squash) > --------------------------------------------------- > Client: (casper) > > --------------------------------------------------- > /etc/fstab > > # <file system> <mount point> <type> <options> <dump> <pass> > none /proc proc defaults 1 0 > > # 192.168.0.1 is 10Mbps > # 192.168.1.1 is 100Mbps > > # first mount at 10Mbps > # <file system> <mount point> <type> <options> > <dump> <pass> > 192.168.0.1:/nfsroot/casper/ / nfs > rsize=8192,wsize=8192,nolock 0 1 > > # then remount at 100Mbps > # <file system> <mount point> <type> <options> > <dump> <pass> > 192.168.1.1:/nfsroot/casper/ / nfs > rsize=8192,wsize=8192,nolock 0 1 > 192.168.1.1:/root/ /root nfs > rsize=8192,wsize=8192,nolock 0 1 > 192.168.1.1:/home/jason/ /home/jason nfs > rsize=8192,wsize=8192,nolock 0 1 > 192.168.1.1:/boot/ /boot nfs > rsize=8192,wsize=8192,nolock 0 1 > 192.168.1.1:/usr/src/ /usr/src nfs > rsize=8192,wsize=8192,nolock 0 1 > 192.168.1.1:/usr/audio/ /usr/audio nfs > rsize=8192,wsize=8192,nolock 0 1 > 192.168.1.1:/usr/video/ /usr/video nfs > rsize=8192,wsize=8192,nolock 0 1 > > # <file system> <mount point> <type> <options> <dump> > <pass> > /dev/cdrom /cdrom iso9660 defaults,ro,user,noauto 0 0 > /dev/dvd /cdrom iso9660 defaults,ro,user,noauto 0 0 > /dev/fd0 /mnt/floppy auto defaults,user,noauto 0 0 > > ------------------------------------------------------- Hmm, I think this is the major problem - I don't think that you're allowed to do such things in fstab because it has no reason to supply the -o remount option when trying to mount the fs the second time so the mount will fail. I would suspect so anyway - I may be wrong. I would therefore suggest that you create a script that is run once the system is up (use /etc/init.d/skeleton as a guide if you want) that i) mount -o remount,rsize=8192,wsize=8192 -t nfs 192.168.1.1/nfsroot/casper / ii) mounts the rest. Really this is still an ugly solution. A more elegant system would go: i) client logs in an grabs a filesystem image which is uncompressed into RAM, all over the 10MBit card. ii) client inflates image, mounts it and boots it (initrd style). iii) client drops 10MBit connection, starts up 100MBit connection, mounts nfs, moves ramdisk off. I would imagine that you have a lot of this already going. The problem is that your setup to too complicated IMHO, simply use the 10MBit card to boot, then drop 10mbps and start everything on the 100mbps. I'm assuming that you're using a boot chip on your 10mbps card to tell it where to get the kernel etc, in which case it should just be a matter of redesigning the initrd image. HTH, Matthew -- Matthew Sackman Nottingham England BOFH Excuse Board: Bogon emissions
pgpJJoJydmE6n.pgp
Description: PGP signature
